Blackwell handbooks of developmenal psychology erika hoff marilyn s...
Upcoming SlideShare
Loading in...5

Blackwell handbooks of developmenal psychology erika hoff marilyn s...






Total Views
Views on SlideShare
Embed Views



0 Embeds 0

No embeds



Upload Details

Uploaded via as Adobe PDF

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
Post Comment
Edit your comment

    Blackwell handbooks of developmenal psychology erika hoff marilyn s... Blackwell handbooks of developmenal psychology erika hoff marilyn s... Document Transcript

    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Search… Submit Query Upload Browse   Go PRO Login Signup Email Favorite Save file   Flag Embed Related More Upload your own Unit11mindmap Chapter 9 outline Language acquisition2 Santrock tls 5_ppt_ch09 Blackwell Handbooks of Developmenal Psychology 222 Language acquisition2 Erika Hoff Marilyn Shatz (editors) Blackwell Handbook views of Language Development (Blackwell Handbooks of + Follow Developmental Psychology) - Blackwell Publishing L by Imbang J. Trenggana on Apr 16, 2012 Theories of language acquisition Developmenal Psychology More… More… Theories%20of%20 Language%20 Acquisition[1] No comments yet Theories Of Language Acquisition[1] Theories Of Language Acquisition[1][9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Subscribe to Chomsky2 Post Comment comments Chomsky2 Bfla0 (With Notes) Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn Shatz (editors) Blackwell Handbook of Language Development (Blackwell Handbooks of Developmental Assignment On Approaches To Language Acquisition Psychology) - Blackwell Publishing L — Document Transcript 1. Blackwell Handbook of Language Development Assignment On Approaches To Language Acquisition 2. Blackwell Handbooks of Developmental PsychologyThis outstanding series of handbooks provides a cutting-edge overview of classic research,current research and future trends in developmental psychology.• Each handbook Approaches To Language Acquisition draws together 25–30 newly commissioned chapters to provide a comprehensive overview of a sub-discipline of developmental psychology.• The international team of contributors to each handbook has been specially Chapter 9 chosen for its expertise and knowledge of each field.• Each handbook is introduced and contextualized by leading figures in the field, lending coherence and authority to each volume.The Blackwell Handbooks of First language acquisition Developmental Psychology will provide an invaluable overviewfor advanced students of developmental psychology and for researchers as an authorita- tive definition of their chosen field.PublishedBlackwell Handbook of Infant Psy i ch.9 DevelopmentEdited by Gavin Bremner and Alan FogelBlackwell Handbook of Childhood Social DevelopmentEdited by Peter K. Smith and Craig H. HartBlackwell Handbook of Childhood Cognitive DevelopmentEdited by Usha Chomskyan linguistics 2 GoswamiBlackwell Handbook of AdolescenceEdited by Gerald R. Adams and Michael D. BerzonskyThe Science of Reading: A HandbookEdited by Margaret J. Snowling and Charles HulmeBlackwell Handbook of Early Chomskyan linguistics 2 Childhood DevelopmentEdited by Kathleen McCartney and Deborah A. PhillipsBlackwell Handbook of Language DevelopmentEdited by Erika Hoff and Marilyn Shatz 3. Blackwell Handbook ofLanguage DevelopmentEdited byErika Hoff and Marilyn Shatz 4. © 2007 by Blackwell Publishing LtdBLACKWELL PUBLISHING350 Main Street, Malden, MA 02148-5020, USA9600 Garsington Road, Oxford OX4 2DQ, UK550 Swanston Street, Carlton, Victoria 3053, AustraliaThe right of Erika Hoff and Marilyn Shatz to be identified as the Authors of the EditorialMaterial in this Work has been asserted in accordance with the UK[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Copyright, Designs, andPatents Act 1988.All rights reserved. No part of this publication may be reproduced, stored in a retrieval system,or transmitted, in any form or by any means, electronic, mechanical, photocopying, recordingor otherwise, except as permitted by the UK Copyright, Designs, and Patents Act 1988,without the prior permission of the publisher.First published 2007 by Blackwell Publishing Ltd1 2007Library of Congress Cataloging-in-Publication Data Blackwell handbook of language development/edited by Erika Hoff and Marilyn Shatz. p. cm. (Blackwell handbooks of developmental psychology) Includes bibliographical references and index. ISBN-13: 978-1-4051-3253-4 (hardcover : alk. paper) ISBN-10: 1-4051-3253-1 (hardcover : alk. paper) 1. Language acquisition. I. Hoff, Erika, 1951– II. Shatz, Marilyn. P118.B583 2007 401′.93—dc22 2006028202A catalogue record for this title is available from the British Library.Set in 10.5 on 12.5 pt Adobe Garamondby SNP Best- set Typesetter Ltd, Hong KongPrinted and bound in Singaporeby Markono Print Media Pte LtdThe publisher’s policy is to use permanent paper from mills that operate a sustainable forestrypolicy, and which has been manufactured from pulp processed using acid-free and elementarychlorine- free practices. Furthermore, the publisher ensures that the text paper and cover boardused have met acceptable environmental accreditation standards.For further information onBlackwell Publishing, visit our 5. ContentsList of Contributors viiiPreface xi 1. On the Development of the Field of Language Development 1 Marilyn ShatzPart I. Basic Foundations and Theoretical Approaches toLanguage Development 17 Introduction 19 2. The Neurodevelopmental Bases of Language 21 Valerie L. Shafer and Karen Garrido-Nag 3. Formal and Computational Constraints on Language Development 46 Helen Goodluck 4. Domain-General Learning Capacities 68 Jenny R. Saffran and Erik D. Thiessen 5. How Inherently Social is Language? 87 Dare Baldwin and Meredith Meyer 6. Input and the Acquisition of Language: Three Questions 107 Virginia C. Mueller Gathercole and Erika Hoff 6. vi Contents 7. The Emergence of Language: A Dynamical Systems Account 128 Julia L. EvansPart II. Language Development in Infancy 149 Introduction 151 8. Experiential Influences on Speech Perception and Speech Production in Infancy 153 Linda Polka, Susan Rvachew, and Karen Mattock 9. Acquiring Linguistic Structure 173 LouAnn Gerken10. Cognitive Processes in Early Word Learning 191 Diane Poulin-Dubois and Susan A. Graham11. Syntactic Supports for Word Learning 212 Letitia R. Naigles and Lauren D. SwensenPart III. Language Development in Early Childhood 233 Introduction 23512. Phonological Development 238 Carol Stoel-Gammon and Anna Vogel Sosa13. Mechanisms of Word Learning 257 Gil Diesendruck14. The Abstract Nature of Syntactic Representations: Consequences for a Theory of Learning 277 Jeffrey Lidz15. Conversational Understanding in Young Children 304 Michael Siegal and Luca Surian16. Bilingual First Language Acquisition 324 Fred Genesee and Elena NicoladisPart IV. Language Development after Early[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Childhood 343 Introduction 34517. Developing Linguistic Knowledge and Language Use Across Adolescence 347 Ruth A. Berman 7. Contents vii18. Language and Literacy in Bilingual Children in the Early School Years 368 D. Kimbrough Oller and Linda Jarmulowicz19. Second Language Acquisition in Childhood 387 Johanne ParadisPart V. Atypical Language Development 407 Introduction 40920. Children with Specific Language Impairment: Bridging the Genetic and Developmental Perspectives 411 Mabel L. Rice21. Atypical Language Development: Autism and Other Neurodevelopmental Disorders 432 Helen Tager-Flusberg22. Reading and Reading Disorders 454 Heikki Lyytinen, Jane Erskine, Mikko Aro, and Ulla RichardsonAuthor Index 475Subject Index 482 8. ContributorsMikko Aro is a researcher at the Niilo Mäki Institute, Jyväskylä, Finland.Dare Baldwin is Professor of Psychology at the University of Oregon, Eugene, Oregon,USA.Ruth A. Berman is Professor Emeritus in the Department of Linguistics at Tel AvivUniversity, Tel Aviv, Israel.Gil Diesendruck is Senior Lecturer of Psychology and Member of the Gonda BrainResearch Center at Bar-Ilan University, Ramat-Gan, Israel.Jane Erskine is a researcher at the Niilo Mäki Institute, Jyväskylä, Finland.Julia L. Evans is Associate Professor at the School of Speech, Language, and HearingSciences at San Diego State University, San Diego, California, USA.Karen Garrido-Nag is a doctoral candidate at The Graduate School and UniversityCenter, City University of New York, New York, USA.Virginia C. Mueller Gathercole is Professor of Psychology at the University of Wales,Bangor, Wales.Fred Genesee is Professor of Psychology at McGill University, Montreal, Canada.LouAnn Gerken is Professor of Psychology and Linguistics and Director of theCognitive Science Program at the University of Arizona, Tucson, Arizona, USA.Susan A. Graham is Associate Professor of Psychology at the University of Calgary,Calgary, Canada.Helen Goodluck is Anniversary Chair of Linguistics at the University of York, York,UK. 9. List of Contributors ixErika Hoff is Professor of Psychology at Florida Atlantic University, Davie, Florida,USA.Linda Jarmulowicz is Assistant Professor of Audiology and Speech-Language Pathologyat the University of Memphis, Memphis, Tennessee, USA.Heikki Lyytinen is Professor of Developmental Neuropsychology at the University ofJyväskylä, Jyväskylä, Finland.Jeffrey Lidz is Associate Professor of Linguistics at the University of Maryland, CollegePark, Maryland, USA.Karen Mattock is a postdoctoral fellow in the McGill Centre for Language, Mind andBrain, Montreal, Canada.Meredith Meyer is a doctoral student in Psychology at the University of Oregon,Eugene, Oregon, USA.Letitia R. Naigles is Professor of Psychology at the University of Connecticut, Storrs,Connecticut, USA.Elena Nicoladis is Assistant Professor of Psychology at the University of Alberta,Edmonton, Alberta, Canada.D. Kimbrough Oller is Professor and Plough Chair of Excellence in Audiology andSpeech-Language Pathology at the University of Memphis, Memphis, Tennessee,USA.Johanne Paradis is Associate Professor[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... of Linguistics at the University of Alberta,Edmonton, Alberta, Canada.Linda Polka is Associate Professor of Communication Sciences and Disorders at McGillUniversity, Montreal, Canada.Diane Poulin-Dubois is Professor of Psychology at Concordia University, Montreal,Canada.Mabel L. Rice is Fred and Virginia Merrill Distinguished Professor of Advanced Studiesat the University of Kansas, Lawrence, Kansas, USA.Ulla Richardson is a researcher at the University of Jyväskylä, Jyväskylä, Finland.Susan Rvachew is Assistant Professor of Communication Sciences and Disorders atMcGill University, Montreal, Canada.Jenny R. Saffran is Professor of Psychology at the University of Wisconsin – Madison,Madison, Wisconsin, USA.Valerie L. Shafer is Associate Professor of Speech and Hearing Science at The GraduateSchool and University Center, City University of New York, New York, USA.Marilyn Shatz is Professor of Psychology and Linguistics at the University of Michigan,Ann Arbor, Michigan, USA. 10. x List of ContributorsMichael Siegal is Marie Curie Chair in Psychology at the University of Trieste, Trieste,Italy, and Professor of Psychology at the University of Sheffield, Sheffield, UK.Anna Vogel Sosa is a doctoral student in Speech and Hearing Sciences at the Universityof Washington, Seattle, Washington, USA.Carol Stoel-Gammon is Professor of Speech and Hearing Sciences at the University ofWashington, Seattle, Washington, USA.Luca Surian is Associate Professor of Psychology at the University of Trieste, Trieste,Italy.Lauren D. Swensen is a doctoral student in Psychology at the University of Connecticut,Storrs, Connecticut, USA.Helen Tager-Flusberg is Professor of Anatomy and Neurobiology and Pediatrics atBoston University School of Medicine and Professor of Psychology at Boston University,Boston, Massachusetts, USA.Erik D. Thiessen is Assistant Professor of Psychology at Carnegie Mellon University,Pittsburgh, Pennsylvania, USA. 11. PrefaceAs they watch their children begin to acquire and use language, that benchmark ofhuman behavior, parents voice both delight and wonder. Researchers, too, have expressedwonder at the children’s accomplishments, and although the scientific study of languageacquisition is relatively recent as topics of academic investigation go, they have made ita lively one, marked by often heated debate. Between us, we have observed nine childrenand grandchildren learning to talk, and we have gained much insight from them. Also,taken together, our work over three decades tallies up to more than 50 years of scientificresearch. We have witnessed many developments since one of us, as an assistant professor,advised the other, her first PhD student at the University of Michigan. We workedtogether then, investigating questions of the role of input. Now, more than 25 years later,we have collaborated again to assess changes in the field and to gather chapters into aHandbook for present-day students that conveys the current state of the science in thecontext of the field’s history. From the start, we envisioned a volume organized around periods of developmentinstead of the more conventional organization around the different aspects of language.We aimed with such an[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... organization to emphasize how the field has been changing fromone that focused primarily on the logical problem of language acquisition to one thataddresses not only this issue but more directly considers the development of language inactual children by examining both their ability to learn any language and also their accom-plishments in learning one or more particular ones over time. An introductory chaptersetting recent work broadly in its historical context, evaluating the state of the science, andraising questions for future research helps to keep the organization true to our original aim.It and the chapters that follow offer a contemporary account of the field. Our primary thanks go to our authors, who heeded our requests for state of the artchapters that set topics in historical context, critically summarized the current state ofresearch, and offered suggestions for the future. They graciously dealt with our proddingover deadlines, as well as our advice on revisions. We met several times over the course 12. xii Prefaceof several years to work on the Handbook, sometimes at conferences we both conveni-ently attended, and twice in Florida, where Brett Laursen generously offered us a placeto work face-to-face, a happy change from the endless e-mailings. Funds from theConsortium for Language, Society, and Thought at the University of Michigan facili-tated face-to-face meetings. Erika’s thanks go also to Sarah Bird, who as acquisitions editor for Blackwell firstsuggested undertaking this Handbook, and to Brett Laursen, who shared the benefit ofhis greater experience in the role of editor of a collected volume. Marilyn’s thanks goalso to her many classes of students in her graduate language development classes at theUniversity of Michigan, especially the students in the 2005 class, none of whom werespecialists in language development, but who nevertheless brought deep interest and freshperspectives to the topic. Her colleagues at Michigan, especially Sam Epstein, RickLewis, and Twila Tardif, as well as visitors to the university, particularly Lila Gleitman,Elena Nicoladis, John Trueswell, and Charles Yang, deserve thanks too for informativeseminars and stimulating conversations. Finally, we would both like to express our appreciation to Sarah Bird for her adviceand encouragement during the early stages of the project and to others at BlackwellPublishing who helped in the making of this book: Andrew McAleer, Elizabeth-AnnJohnston, Jenny Phillips, and Simon Eckley. Special thanks are due to Melanie Weiss,of Florida Atlantic University, who ably and cheerfully provided secretarial assistance tothe project. Erika Hoff Marilyn Shatz 13. 1On the Development of the Field ofLanguage DevelopmentMarilyn Shatz “Plus ça change, plus c’est la même chose.” A conversation between a 2- year-old boy and his grandmother (1991): gm: (reading) There were many, many buildings in the city. c: Many, many builds. A conversation between a 31/2 -year-old girl and her grandmother (2006): gm: They’re building a house. c: No, not “building.” Say “making.”Speculation about how children come to know a language antedates by centuries thesystematic investigation of the[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... topic; only in the last century did the methodical studyof language development explode. In roughly 100 years (dating the field’s origins fromthe diaries of the Sterns, 1907) there have been a myriad of studies on children’s languagedevelopment and many consequent changes in both method and theory. Within thisrelatively short history, the field has reflected the changing “hot” intellectual trends ofthe times. Thus, the first half of the twentieth century saw descriptive, normative worklike McCarthy’s (1930), whereas much work in the latter half of the twentieth century,influenced by theoretical work in linguistics, attempted to prove or disprove claims aboutpredispositions specific to language (see Pinker, 1994). Now, in the early twenty-firstcentury, much research in language development is concerned with brain developmentand computational skills, cross-linguistic and cultural comparisons, and bilingualismand education. This Handbook strives to capture the state of the art by bringing togetherchapters written by a generation of researchers who mainly address current concerns. However, a reading of the chapters in this Handbook reveals that questions ofpredispositional specificity, though changed, are still with us. The first argumentsfor language-specific predispositions made an obviously false but simplifying 14. 2 Marilyn Shatzassumption that language acquisition was instantaneous (see Chomsky, 1975). Instead,more recent work makes development an important factor in the language acquisitionstory, thereby requiring renewed evaluation of the roles played in an ongoing acquisitionprocess by the child’s physiological, cognitive, and social status, as well as caregiverinput. There is, then, increased appreciation for how multiple interacting factors maycontribute to language acquisition and how weightings among them may change withdevelopment. Moreover, it is clear that the predisposition issue can be divided into a question aboutwhat the predispositions are and a question about whether any of these are necessarilyspecific to the task of acquiring language. In this volume, there are chapters reportingfindings sometimes relevant to one question and sometimes both. With regard to lan-guage- specific constraints, for example, Diesendruck, in his chapter, asks whether mech-anisms that limit possible word meanings are specific to that domain or are more general,and Lidz, in his chapter, asks whether children’s early syntactic knowledge supports evi-dence of early abstract grammatical representations. Whereas the evidence argues againstdomain-specific predispositions for word learning, it is premature to decide the case forsyntactic competence (see the section Do we still need domain-specific constraints? for morediscussion). Despite the changes the field has undergone in recent years, the continuingconcern about the nature of the child’s abilities is an apt instance of the idea that themore there is change, the more there is constancy. There is other evidence of constancy in the face of change. Four decades ago, system-atic research on language development was almost exclusively focused on how youngchildren acquire English as their native language. More recently, research has expandedto include work on older children and has produced findings on how linguistic[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... knowl-edge develops with experiences such as learning a second language and learning to write,and how earlier developments in oral language are related to the later acquisition ofreading (see chapters by Oller & Jarmulowicz and by Paradis on second-language learn-ing, by Berman on later language use, and by Lyytinen, Erskine, Aro, & Richardson,on learning to read). Researchers have also asked whether and how the course of acquisi-tion changes as a function of cultural differences or differences in language types (see,e.g., Slobin, 1985, 1992, 1997). Work in this vein revealed some interesting facts aboutearly language development. For example, child- directed language is not necessary toacquisition (e.g., Ochs & Schieffelin, 1984), and children are sensitive very early on tolanguage-particular features, even those encoding such basic notions as space, time, andmanner (e.g., Bowerman, 1996; Bowerman & Choi, 2001). Children acquire a languageunder many different conditions of exposure; yet regardless of those conditions, theyseem to make ordered decisions about the particular language to which they are exposed(Baker, 2005). Moreover, children learning the same language in different times and under differentconditions also exhibit similar concerns during development. The quotes above, fromtwo children who were learning English in different decades and rearing conditions,reveal that both children by age 3 were troubled by the potential ambiguity in gram-matical class of the word building, although they apparently had previously analyzed theword in different ways. One seemed concerned that the morphological ending -ing benon-ambiguous in indexing verbs, whereas the other seemed to have coded the word as 15. On the Development of the Field 3a noun, regardless of ending, and refused to accept a usage that would result in anambiguity of grammatical class. Such anecdotes suggest that, at least by age 3, youngchildren do not mindlessly accept the language they hear, but they think about it some-what abstractly and creatively. Thus, while the field has expanded to address a broader range of topics concerninglanguage development and use, it is important to note that the central phenomenonneeding explanation has not much changed – children acquire a great deal of the lan-guage of their community in a comparatively short time without much direct tuitionand with remarkable commonality in the concerns they have. Both the similarities anddifferences in their patterns of development require explanation. Also, it is common, though not often recognized by scientific researchers, that currentstates of knowledge rest on the contributions and efforts of prior generations. In theabstract, science proceeds via the falsification of theories; but in practice, wholesalechanges or substitutions of alternative world views are rare. Instead, increasingly sophis-ticated empirical investigations provide the engine for theoretical change. While debateswithin a field are often portrayed as dichotomous – with positions declared to be eitherformalist or functionalist, black or white, right or wrong, proven or disproven – thereality is often not so discontinuous, and progress in a field typically comes via rap- prochement, as both earlier questions and earlier answers are modified to[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... accommodatemore recent findings. As much as language development research has progressed in recentyears, the footprints of its past are still visible. In this chapter, I elaborate on these observations by addressing some themes of lan-guage development research, taking note of some of the precursors to the more recentwork, and suggesting promising research topics for the future.The Enduring Question about Language DevelopmentEvery human language is a complex system. Each includes regular ways of combining alimited set of signs or sounds to create an unlimited set of meanings. Any relativelynormal child born into a relatively normal language- using community will, even withoutformal education, develop the ability to use that community’s language. How this featis accomplished so readily has been and still is the central question of the field of languagedevelopment.The legacy of theories of syntactic formalismAre areas of language autonomous? The work of two language theorists greatly affectedthe organization of twentieth-century research in the field of language development.First, the philosopher and semiotician Charles Peirce described four distinct componentsof language: phonology, semantics, syntax, and pragmatics. Second, linguists, mostnotably Chomsky, reified these distinctions by offering compelling theories, particularlyof syntax, that assumed for analytic purposes the autonomy of separate parts. Scholarly 16. 4 Marilyn Shatzwork and curricula in linguistics departments came to be organized around the distinc-tions, and even handbooks and textbooks of language development (e.g., Fletcher &MacWhinney, 1995; Hoff, 2005) commonly organized the field around them. Peirce’s distinctions help even today in understanding the complex phenomena oflanguage and its acquisition. After all, children learn the signs or sounds of a language,they learn words and word meanings, they learn how to combine words into meaningfuland acceptable sentences, and they learn how to interpret such sentences in a variety ofcircumstances. A reasonable, scientifically sound, first approach was to analyze all theseaccomplishments separately and to see how far one could go with the simplifyingassumption of autonomy of components. The autonomy approach fueled a lively com-munity of researchers in syntactic theory, and its application to the field of languageacquisition did have some success, perhaps most clearly with regard to the knowledgegained about the capacities of infants to analyze the sound patterns of their language(see Polka, Rvachew, & Mattock, this volume). Yet, the autonomy assumption was made for analytic purposes; it is an empiricalquestion whether children necessarily abide by it when developing language. Increas-ingly, researchers propose interactive theories, where one kind of linguistic knowledgecan influence, or is even crucial to, the development of another (see, e.g., Gleitman,Cassidy, Nappa, Papafragou, & Trueswell, 2005). In addition, formal linguists areincreasingly recognizing that the ways in which the components interface, and eveninteract, are crucially important to analytic characterizations of adult language (e.g.,Chomsky, 1995). To capture this trend, as well as to emphasize the factor of[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... development in the acquisi-tion of language, we chose to organize the sections of this volume according to an agecontinuum rather than by components of language. Although the chapters within eachPart by and large still reflect the fact that researchers focus more on a specific componentof linguistic knowledge than on interactions among components, the growing tendencyto account for the phenomena of one with explanations partly dependent upon anotheris nevertheless well represented. Currently, the topic of word learning is the best exampleof an area in the acquisition field where researchers propose that knowledge of one lan-guage component helps to account for development in another. So, in this volume, achapter by Naigles and Swensen, one by Stoel-Gammon and Sosa, and one by Diesend-ruck argue for syntactic, phonological, and pragmatic knowledge, respectively, as factorsaiding word learning. The idea of interaction across components of language has found more advocates inrecent years, and so has the idea of interaction among areas of development. Currently,infants’ social abilities play a large role in some language development theorists’ thinking(e.g., Baldwin & Meyer, this volume), and the question of the separateness of languageand cognition is receiving renewed interest (see Saffran & Thiessen, this volume). Thereare several ways to achieve more interaction among areas of development. Basing lan-guage acquisition on cognitive or social knowledge (e.g., Tomasello, 1992) is one way,but that has been argued to be troubled (see Poulin-Dubois & Graham, this volume;Shatz, 1992). Another way is to integrate language intimately into the very fabric of asystem of development and thereby provide a more unified account of how the childmoves toward adulthood (Shatz, 1994, in press). While the question of how children 17. On the Development of the Field 5acquire the complex system of language is still central, an increasingly common goalamong researchers in the field of language development is to take more account of otherareas of development; theoretical accounts of acquisition that recognize neither a broaderpicture of language development nor the child’s ability to operate across areas are waningin influence.The poverty of the stimulus. Another argument about acquisition that was generated bythe dominance of work in syntactic theory was the one from the poverty of the stimulus(see Chomsky, 1980). The claim here was that infants and children do not hear languagewhich sufficiently or directly reveals to them the underlying and abstract grammaticalstructure governing all and only the grammatical sentences of a particular language.Without such information, children cannot converge on the correct structure of thelanguage they do hear. The idea that language acquisition was constrained by innateuniversal grammatical principles guiding their acquisition filled this input gap. Early controversy over the poverty of the stimulus argument concerned whether theenvironment or the child provided the engine of development (see Shatz, 1987, for earlyarguments in favor of the child). Decades of research on the language spoken to thechild, beginning in the 1970s, attempted to find evidence against the poverty of thestimulus argument, first by showing that child-directed[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... speech was largely grammatical,and then by proposing that it was richer in structure than earlier envisioned (see Baldwin& Meyer; Gathercole & Hoff; Gerken, this volume, for more history on this topic). Recently, the arguments against the poverty of the stimulus claim have been changingin an important way. Input theorists no longer argue simply for a much richer overt inputto a relatively passive child. Rather, the argument now is that the child has generalcapacities for extracting over time from a range of input sentences a deeper set of rela-tions than are revealed superficially in the input (see, e.g., Elman, 2003). Thus, that thechild provides the impetus for acquisition is more readily accepted, but the characteriza-tion of the child is also changing, from a child endowed simply with an abstract blueprintfor grammar to one with a broader complement of sensitivities and capacities applicableto the language development task. Nonetheless, the critical issue remains how to characterize the capacities the childneeds to traverse the developmental landscape efficiently (see Baker, 2005). Whateverthose capacities, the linguistic environment will have to be sufficient in both variety andquantity to offer the child so equipped what it is she needs to abstract her particularlanguage’s structure; and there are consequences for the rate and quality of developmentif it is not (see Hoff, 1999, 2006). Still controversial are questions of whether any of thecapacities are domain-specific and which are innate or emergent during development.On the one hand, some researchers have provided evidence that children have generalabilities for extracting information from input (see Gerken; Saffran & Thiessen, thisvolume). On the other hand, some researchers offer particular acquisition phenomenaas evidence of the continuing need for language-specific constraints (see Lidz, thisvolume). At least as yet, there are no explicit general accounts of these (see also Lidz &Gleitman, 2004), although there are those who believe that such general capacitiesultimately will account for all of language development (e.g., MacWhinney, 2004;Tomasello, 2003; but see Hoff, 2004; Legate & Yang, 2002). 18. 6 Marilyn ShatzNo negative evidence. Related to the poverty of the stimulus argument is the no-negative-evidence-in-input argument. If, when children make mistakes of grammar asthey acquire language, they neither get corrected nor pay much attention to corrections,how do they recover from their mistakes? One early proposed solution to the no-negative-evidence problem was the subset principle: Children are constrained to considerfirst only those grammars that generate a subset of grammatical sentences so that uponhearing in input the positive evidence of the broader set of sentences, they can adjusttheir grammars (Berwick, 1986). Alternatively, a generation of researchers sought to findevidence against the whole idea of the no-negative- evidence problem by showing thatchildren had at least probabilistic information from parental responses as to the gram-maticality of their utterances, but such efforts proved to be inadequate (see Marcus,1993). In this volume, Goodluck does not eliminate the no-negative-evidence problem.Instead, she tries to avoid Berwick’s otherwise unmotivated subset[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... principle by arguingfor the possibility that the processing system responsible for comprehending and produc-ing language itself provides a developmental constraint on the workings of the linguisticsystem. Namely, limited processing ability constrains access to universal grammar, andthis accounts for errors and omissions under a variety of circumstances. As is made clearin the next section, how processing constraints are construed can have deep implicationsfor a basic tenet of formalist theory.Competence and performance. Obviously, scientific study requires that the phenomenaunder investigation be identified. Chomsky identified his topic of linguistic interest notas human language use, but as the knowledge of grammar adults possess; and the evi- dence of the nature of that knowledge was to be found not in utterances but in judgmentsof grammaticality. In consequence, what is known became divorced from what is said,and errors of speech were argued to be subject to chance factors – vagaries of the momentthat might afflict any speaker at any time, and of no relevance to questions of linguisticcompetence or knowledge. With regard to language development, the issue of how toassess competence and not just performance was a particularly knotty one, since nomatter how clever the experimenters, they still were left with only overt behaviors bywhich to assess underlying knowledge. With very young children, there are no oppor-tunities to obtain judgments of grammaticality. Unsurprisingly, then, the latter half ofthe twentieth century saw many controversies about what constituted evidence for oragainst linguistic competence and what could be discounted as reflecting “only” perfor-mance variables. Recent work in psycholinguistics has sought to bridge the gap between competenceand performance. If, for example, a language processor is constrained by limitations ofhuman working memory, that processor itself may have an impact on the very natureof syntactic structure (see Lewis & Vasishth, 2005). Such a processor apparently differsfrom Goodluck’s proposal because it influences the very nature of language structureitself and not just access to it. Its effects, then, would be constant, not developmental.Possibly there are several sorts of processing constraints, some accounting for structuralcharacteristics and others for how children at different levels of development handlevarious kinds of task demands (see Werker & Curtin, 2005, for an example of the latter).In any case, what seemed like a clear formalist distinction between competence and 19. On the Development of the Field 7performance has become blurred. Indeed, doing away altogether with the distinction isthe radical solution proposed by dynamic systems theorists (who do away with the stan-dard notion of representation as well). By integrating knowledge and performance infavor of real-time processing, Evans argues in this volume that language is “performancein context.” There is still the question, however, of just what is meant by “processing.” Is there adomain-specific processor that affects the very structure of language? Is there as well amore general processor with additional limitations that also affects processing of linguis-tic as well as non- linguistic information? Are there unique limitations on processing dueto[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... immaturity? What is the relation between processing and representation? The increasedinterest in “processing” – and the attendant demise of the competence–performancedistinction – is indicative of the promise of processing arguments, although such argu-ments are in need of more empirical support and more theoretical clarification.What constrains the acquisition of language?In 1965, Chomsky proposed a language acquisition device (largely for syntax) that wasendowed with a universal grammar such that it received specific-language input whichwas underdetermined with regard to that language’s underlying structure but nonethelesscould instantly output any and only grammatical sentences of that language. The uni-versal grammar itself consisted of both constraints on allowable human languages – whatthey could not be – and specifications of what they could consist of, thereby limitingthe allowable conclusions the device could draw from the input received about the natureof the particular language to be learned. We knew in 1965 that the device had to be flexible enough to acquire any humanlanguage. As a result of subsequent research on the acquisition of sign languages, wenow know that the device is not limited even to one modality (e.g., Petitto & Marentette,1991); and as a result of research on bilingual development, we know that it is not limitedto acquiring one language at a time (Genesee & Nicoladis, this volume). Also, by apply-ing new genetic and neurobiological techniques to the study of language develop-ment, we have discovered that various atypical outcomes can arise from geneticallybased, neurodevelopmental disorders (Rice; Shafer & Garrido-Nag; Tager-Flusberg, thisvolume). We have also seen that the question of what constrains the acquisition of languagereceives a much broader answer today. Not only do researchers in language developmentgenerally accept a broader scope to the field, but they also accept the fact that languageacquisition is not instantaneous but is subject to development, with the weightings ofvarious influential factors changing as the child’s knowledge states change. Interactionsamong components of language knowledge, as well as among different areas of develop-ment – in particular the physiological, social, and cognitive spheres – are increasinglyrecognized as relevant to the complex process of language acquisition. Currently, the“language acquisition device” can be described as a probabilistic thinker, capable ofmaking inductive inferences across utterances over time based on a broad array of cueschanging with development. 20. 8 Marilyn Shatz In the next section, I consider some of the trends in past research that foreshadowedthe current, more comprehensive, view of the language learner, and I ask whether thecurrent view suffices to do without the idea of domain-specific abilities.The Nature of the Child Developing LanguageThe active, social childThe field of language development can thank Jean Piaget for the notion of a child activelynavigating a course of development. Beyond that important but general view, however,Piaget’s characterization of a child progressing through stages of cognitive developmentgoverned by dialectical mechanisms of assimilation and[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... accommodation offered littlehelp in explaining the phenomena of language development. Indeed, attempts to findcases in which language milestones could be based directly on cognitive developmentsbore relatively little fruit (see Poulin-Dubois & Graham, this volume, for discussion).Bruner’s attempts to base language development on the child’s active social interactionssimilarly met with failure (see Baldwin & Meyer, this volume, for a review). It is notbecause language development is autonomous that such attempts failed. Rather, it isbecause they did not recognize how integrated these areas of development are for thechild; the divisions of social, cognitive, and language development are divisions of theresearcher, not of the child. The task of a developing child is to become an active,knowledgeable participant of a language community. To do so, she uses whatever pre-dispositions and knowledge she has to increase her understanding even as she engagesin it. It is in this sense that she bootstraps her way to adult competence as an active,social child (Shatz, 1987, 1994). Two decades ago, I summarized the evidence available then for active child languagelearners who “use what they have to learn more” (p. 1), and I proposed a set of capacitiesa normal language learner would use to control the process of acquiring a native language(Shatz, 1987). The capacity to elicit speech in referentially transparent situations throughgesture and eye gaze, and eventually, simple utterances, was an ability that I suggestedserved the function of eliciting data potentially rich, not only in syntactic information,but also in pragmatic and semantic information (“elicitation operations”). Second, Isuggested that the capacity to store strings in units large enough for off-line comparisonand analysis was necessary (“entry operations”). Finally, I proposed a set of “expansionoperations” that provided the child the opportunity to explore and analyze what sheknew of language – its structure and use – through language play, practice, and organi-zation, thereby to generalize and self- repair. Much of what appears in this volume is not only compatible with these proposals,but supports, elaborates, and clarifies them. New findings on children’s statistical learn-ing abilities (Saffran & Thiessen, this volume) and sensitivity to form (Gerken, thisvolume) give added specification and substance to the ideas of entry and expansionoperations. Evidence for pragmatic and syntactic bootstrapping (see Diesendruck; Naigles& Swensen, this volume) is consonant with the claim that children use what they know 21. On the Development of the Field 9to learn more, and so is evidence that children, by virtue of their immaturity, inexperi-ence, or incomplete knowledge, sometimes make inaccurate syntactic self-repairs (Shatz& Ebeling, 1991) or inappropriate pragmatic inferences (Siegal & Surian, this volume).Research on language disorders also supports and expands on my argument that differentdisorders result from the disruption of different kinds of operations the child can carryout (see Rice; Tager-Flusberg, this volume). An active child, then, is one who brings avariety of capacities to the complex task of language acquisition, and based on whateverher capacities and knowledge at the time, she makes inferences about her linguistic situ-ation at that moment. What role[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... remains for the environment? In a comprehensive review of the influencesof variable social environments on language development, Hoff (2006) provides consid-erable evidence for the myriad social factors that affect development, from the amountof time children spend in conversation to the contexts in which they hear talk. Hoffproposes that, regardless of differences in such social circumstances, all environmentssupportive of language must provide two things: opportunities for communicative inter-action and an analyzable language model. However, as Hoff notes, no matter how richan environment is in such opportunities, it cannot by itself be effective as a support forlanguage development if the child does not take advantage of them. To optimize their learning opportunities, then, children have to be social beings,interested and able to engage in communicative interactions involving language, verypossibly using the proposed elicitation operations to maintain and enhance the learningenvironment, as well as having the skills to analyze the language model it offers. Howmuch opportunity they have to exercise their skills, and how well their language modelmatches their abilities for analysis will have an effect on the rate and course of theirparticular language development. None of this seems especially arguable. Infants typically engage with their interlocu-tors early and often; infants and toddlers who don’t do so have problems with languagedevelopment, sometimes very severe ones (see Baldwin & Meyer; Tager-Flusberg, thisvolume). Yet, even some children who don’t engage normally can acquire a fair amountof language, including a vocabulary of reasonable size (Diesendruck, this volume). Howsociable does a child have to be to acquire any language at all? If indeed children use what they know to learn more, one possibility is thatchildren who are less sociable substitute other skills when learning language. Forexample, when learning new words, children with autism may make few inferences (ornone at all) about the speaker intentions that apparently inform typically developingchildren about word meaning; rather, they may focus more on associations or distribu-tional cues. This view fits well with the argument that there are multiple clues to thenature of language. Individual differences in quality and rate of acquisition woulddepend, then, not only on differences in social environments but also on which clueschildren with varying skills could use in varying environments to access them.Social– linguistic environments need to provide multiple cues a child can use to discoverhow to act and interact both linguistically and non-linguistically in her community. Butthe developing child actively and selectively attends to the cues as she grows in bothbrain (Shafer & Garrido-Nag, this volume) and mind (Poulin-Dubois & Graham, thisvolume). 22. 10 Marilyn ShatzDo we still need domain-specific constraints? Conversations as information conduits. Infants have a preference for speech sounds overnon-speech sounds. This attentional bias toward language suggests that the human childmay be prepared to engage with conspecifics via language, the benchmark behavior ofhumans. Such a bias may be the initial entrée into developing one’s own language com-petence. Also, it may[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... facilitate a very early infant behavior – engaging in reciprocalinteractions involving vocalizations, which is typically rewarding for both caregiver andinfant. Such “proto-conversations” become increasingly sophisticated, and conversationsare major sources of information for the developing child (see Shatz, in press; Siegal &Surian, this volume). There are many ways young children can learn – by observingothers, or by trial and error, for example. But no ways other than conversation provideinformation via indirect experience, or what Harris (2002) has called “testimony.” More-over, the richness and variety of information that can be conveyed in relatively shorttime is unmatched by other means. In short, engaging in conversations via language isa powerful tool for learning and not just learning language. So far, the picture we have of the optimally successful language learner is one of anactive, social child equipped with early attention to language, possessing probabilisticinferencing skills, and engaged in frequent communicative interactions rich with lin-guistic information. Can a child so equipped traverse the linguistic landscape efficientlyenough to learn the substance of any of thousands of human languages by school age?Language typologies as a constraint on the learning task. Whether the learning problem,especially with regard to the acquisition of syntax, can be settled solely by recourse tothe kind of child described above is far from clear (see Baldwin & Meyer, this volume,for more discussion). Chomsky’s early insights led him to argue for an innate universalgrammar based on abstract principles underlying all human languages. A later proposalclarified how both the commonalities and the differences among languages could beaccommodated in such a system: All human language can be described in terms of asmall number of dimensions or parameters having a limited number of settings, witheach setting potentially accounting for apparently disparate superficial characteristics.Any specific language, then, can be described by its particular combination of set values;and the vast numbers of languages result from the various possible combinations of set-tings on those parameters (see Chomsky, 1981). According to this view, the learningtask is more manageable because, rather than considering a virtually unlimited numberof hypotheses, a child equipped with knowledge of that small number of parameterswould have only to determine which of the possible parametric values pertain to theparticular language to be learned. Despite the elegance of this approach, attempts toapply it directly to account for the data of language acquisition have been controversialat best (see Goodluck, this volume, for more on parameters and their application to theproblem of language acquisition). Baker (2001, 2005) has offered a version of parameter setting that seems more com-patible both with recent findings on universals of grammar and, at least on the surface,with the picture of the child as a probabilistic thinker who learns language over time,keeping track of correlational properties and making inferences from them. Rather than 23. On the Development of the Field 11couching parameters in terms of formal universal principles, Baker bases them on gen-eralizations from[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... implicational universals. These are descriptive statements about lan-guage commonalities and differences with respect to the ways linguistic propertiescorrelate with one another, and they are stated in terms of probabilities; for example,70% of languages do one thing and not another. He notes that some parameters canhave a greater impact on a particular language’s form than others, and he orders theparameters hierarchically according to their ability to affect one another; parametershigher in the hierarchy than others have at least one setting which eliminates the needto consider other parameters at all. If children first set the parameters higher in the ordering, they would not, dependingon the language they were learning, necessarily need to consider lower-ordered ones atall, thereby narrowing still further the learning task. If the language to be learned didrequire further parameter setting, the errors children made would be consonant withnot yet having considered parameters lower in the hierarchy. Baker’s ideal learner is onewho, in developing a grammar, observes the ordering. Baker provides some preliminaryevidence from children’s errors at early stages of grammatical development showing thatthey do appear to observe that ordering. Baker explicitly leaves open the question of whether the particular learning mecha-nisms needed to account for what he proposes are domain- general or domain-specific.In either case, the particular-language input, or what Baker refers to as the “primarylinguistic data” (p. 119, footnote 9), must contain the information the child needs so asto learn according to the ordering. This requires, then, a symbiosis between the inputand the child’s learning abilities. Recent work on sensitivity to the frequency of linguisticinformation and its consistency in the input suggests that frequency and consistencymonitoring are within the ken of young children (see Gathercole & Hoff, this volume).If different types of information in the input vary enough on these dimensions, the inputmay “order” appropriately the apparent parametric decisions children seem to make. Thereported ordering evidence may be explainable, then, as a consequence of a child, sensi-tive to the frequency and consistency with which the input displays particular phenom-ena, repeatedly operating over such input and inferring structural relationships in amanner consistent with the hierarchy. This sort of explanation helps to dispense withany need for prior knowledge of the parametric hierarchy. However, before one can conclude that domain-specific knowledge is not required,one must ask what the entities themselves are that are being monitored for frequencyand consistency. Are they abstract linguistic concepts, and if so, how can the abstractnessbe accounted for? Implicational universals are typically written in terms of abstractgrammatical entities like verb and direct object, and even the most general descriptionof them, as in Baker’s parameters, utilizes some linguistic abstractions like word orphrase. Baker’s solution does not seem neutral with regard to the source or kind of rep-resentational abilities needed to deal with the abstractions. Baker suggests that childrenhave “an innate phrase building mechanism, which stands ready to group whatever wordsthe child happens to learn into useful[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... larger phrases” (Baker, 2005, p. 104). So, hisproposal relating typological data and acquisition relies on an innate structure-buildingdevice handling abstract linguistic representations. In this regard, it seems little differentfrom and hardly more plausible than earlier claims for innate, language- specificabilities. 24. 12 Marilyn Shatz Despite the issue of their implausibility, is it nonetheless necessary to grant innatedomain-specific capacities to the learner? A “yes” answer may be premature. Saffran andThiessen (this volume) make several important arguments relevant to the question. First,they note that questions of innateness and domain-specificity are orthogonal issues:Logically, either kind of abilities, general or specific, can be innate or emergent. Second,they offer evidence that “domain-general learning abilities” should not be construed ascompletely unconstrained; rather, they are limited by specifically human perceptual andcognitive constraints, and those constrained abilities can shape a specific domain likelanguage to include just those properties (such as phrase-based structures) needed tomesh with the learning capacities. When coupled with the notion of development inwhich such abilities operate repeatedly over inputs themselves previously modified byvirtue of cycles of increasingly sophisticated organizations, this proposal suggests apowerful mechanism, able to account for the emergence in the learner’s competence oflinguistic notions like “verb” or “phrase” (see Saffran & Thiessen, this volume). Thesymbiotic relationship between language typology and acquisition may thereby be pre-served without the need to stipulate innate, domain-specific abilities to account fordomain-specific representations. However, at this point in the course of research into language development, the ques-tions of whether and how a resolution to the domain-specificity debate will bring uscloser to a true understanding of language development remain unanswered. Becausethe constraints on human domain-general abilities are not fully understood, it is as yetunclear how substantive are the differences between the domain-general view and thedomain-specific view. The commonalities among languages attest to the existence ofconstraints somewhere, but whether to situate them in language because of the natureof the learner or in the learner because of the nature of language may turn out simplyto be a question of perspective, reminiscent of the controversy between proponents ofChomskyan and Gibsonian theories about whether constraints on our view of the worldare situated in our minds or in the world. The unhappy possibility here is that this ques-tion is merely a metaphysical one, not amenable to serious empirical inquiry capable ofresolving the issue. The hope that it is not rests on the expectation that further empiricalinvestigations into constraints on domain-general abilities will reveal that their differ-ences from a domain-specific view are not merely superficial. If the differences are sub-stantive, then how the views weigh in on two scientific criteria should help to decide infavor of one or the other: namely, whether one view accounts for more phenomena thanthe other, and whether the explanatory constructs offered by one rather than the otherare more compatible with[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... constructs at other levels of explanation. The goal of domain-general theorists seems to be to win the day on both those counts.Tasks for Future ResearchMany of the authors of chapters in this volume note gaps in our current understandingof language development and make specific suggestions for future research. Here I noteseveral general concerns that have emerged in this review. 25. On the Development of the Field 13 Above all, we need an explanatory theory of language development in whichlanguage acquisition neither stands alone nor is said simply to result from otherkinds of development or knowledge – cognitive or social. If we take seriously theclaim of interactions among areas of development, then language knowledge cannotsimply emerge out of, say, social development, although various aspects of socialunderstanding, of intention and mind, can come into play to expand languageknowledge. We need to expand our understanding of how various areas of developmentinteract mutually to sustain development, and we need as well to continue to addresshow areas within language knowledge impact development of one another. Onlythen can we expect to have a truly comprehensive, explanatory theory of languagedevelopment. Another obvious goal for the future is a more complete understanding of humaninformation processing abilities. We need to know how these human abilities differ fromthose of other creatures, what unique characteristics ours may have, whether and howthey change with development, and how human processing characteristics may influencethe very nature of human languages. We also need to know more about the genetic basis of human abilities. Surely, increas-ing knowledge about how genes work will help us to discover more about the geneticbases of language, but already our growing understanding of human genetics suggeststhat the idea of a single “language gene” is a journalistic fantasy. Genes multi-determinebehavior, including language behavior, and deficits in single genes typically have multipleconsequences, some on the linguistic system, some not. Increasingly, efforts to dissociatelanguage from cognition seem as infelicitous as earlier efforts to correlate languagemilestones with Piagetian cognitive developmental ones. Our knowledge of the development of the human brain is itself in an infant stage.We need to know more about the developing human brain, and to have more evidenceon the limits brain immaturity imposes on the functional ability to extract informationfrom the environment. We also need to know whether, when, and how different aspectsof the environment, such as the availability of more than one input language, have animpact on the developing brain. As we become more knowledgeable in these areas, we will gain even more understand-ing than we now have about how the predispositions that human infants bring to thetask of developing language interface with their language communities to producemature language knowledge. Recognizing their intimate relationship, we nowadays paylittle attention to the nature–nurture dichotomy that for so long characterized debatesin the field of language acquisition. In the future, we may very well discard the domain-general[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... versus domain-specific dichotomy, as we learn more about how human informa-tion processing capacities are innately constrained and how they in turn constrain thestructure of language. We may even decide that the best characterization of the courseof language development is that there is no single course, but instead multiple pathsthrough varying epigenetic landscapes determined by a particularly endowed childacquiring a specific language in a specific cultural context. Still, what constrains thenumber and kinds of those paths will be the limited ways a genetic program interactingwith an “expected” environment in order to build a functional human adult can bemodified by accidents of nature or nurture. 26. 14 Marilyn Shatz The chapters in this book represent the state of the art early in the second century ofsystematic work on the question of how language is acquired by human infants. Theysummarize the history of the field and its progress to date. To their credit, regardless oftheir theoretical bents, the authors have by and large eschewed an antagonistic tone infavor of presenting and interpreting data reasonably even-handedly, recognizing gaps,and noting continuing challenges. Such open-minded candor itself presages well for afield with very old questions and still-developing answers.NoteI owe much to Sam Epstein and Richard Lewis for a provocative seminar on competenceand performance at the University of Michigan in fall, 2004, and to the authors with whomI discussed revisions to their chapters. I also thank Erika Hoff and Richard Feingold forcomments on earlier versions of this chapter. Of course, remaining errors and interpretationsare mine.ReferencesBaker, M. (2001). The atoms of language. New York: Basic Books.Baker, M. (2005). Mapping the terrain of language learning. Language Learning and Develop- ment, 1, 93–129.Berwick, R. (1986). The acquisition of syntactic knowledge. Cambridge, MA: MIT Press.Bowerman, M. (1996). Learning how to structure space for language: A crosslinguistic perspec- tive. In P. Bloom & M. A. Peterson (Eds.), Language and space: Language, speech, and commu- nication (pp. 385–436). Cambridge, MA: MIT Press.Bowerman, M., & Choi, S. (2001). Shaping meaning for language: Universal and language spe- cific in the acquisition of spatial semantic categories. In M. Bowerman & S. Levinson (Eds.), Language acquisition and conceptual development (pp. 475–511). Cambridge: Cambridge Uni- versity Press.Chomsky, N. (1965). Aspects of the theory of syntax. Cambridge, MA: MIT Press.Chomsky, N. (1975). Reflections on language. New York: Pantheon Books.Chomsky, N. (1980). Rules and representations. London: Basil Blackwell.Chomsky, N. (1981). Lectures on government and binding. Dordrecht, The Netherlands: Foris.Chomsky, N. (1995). The minimalist program. Cambridge, MA: MIT Press.Elman, J. (2003). Generalizations from sparse input. In Proceedings of the 38th Annual Meeting of the Chicago Linguistic Society. Chicago: University of Chicago Press.Fletcher, P., & MacWhinney, B. (1995). Handbook of language acquisition. London: Blackwell.Gleitman, L. R., Cassidy, K., Nappa, R., Papafragou, A., & Trueswell, J. C. (2005). Hard words. Language Learning and Development, 1,[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... 23–64.Harris, P. (2002). What do children learn from testimony? In P. Carruthers, S. Stich, & M. Siegal (Eds.), The cognitive basis of science (pp. 316–334). Cambridge: Cambridge University Press.Hoff, E. (1999). Formalism or functionalism? Evidence from the study of language development. In M. Darnell, E. Moravscik, M. Noonan, F. Newmeyer, & K. Wheatlery (Eds.), Functional- ism and formalism in linguistics (pp. 317–340). Amsterdam: John Benjamins. 27. On the Development of the Field 15Hoff, E. (2004). Progress, but not a full solution to the logical problem of language acquisition. Journal of Child Language, 31, 923–926.Hoff, E. (2005). Language development. Belmont, CA: Thomson Wadsworth.Hoff, E. (2006). How social contexts support and shape language development. Developmental Review, 26, 55–88.Legate, J. A., & Yang, C. D. (2002). Empirical reassessment of stimulus poverty arguments. The Linguistic Review, 19, 151–162.Lewis, R., & Vasishth, S. (2005). An activation-based model of sentence processing as skilled memory retrieval. Cognitive Science, 29, 375–419.Lidz, J., & Gleitman, L. R. (2004). Yes, we still need a universal grammar. Cognition, 94, 85–93.MacWhinney, B. (2004). A multiple process solution to the logical problem of language acquisi- tion. Journal of Child Language, 31, 883–914.Marcus, G. F. (1993). Negative evidence in language acquisition. Cognition, 46, 53–85.McCarthy, D. (1930). The language development of the preschool child. Institute of Child Welfare Monograph (Serial No. 4). Minneapolis, MN: University of Minnesota Press.Ochs, E., & Schieffelin, B. B. (1984). Language acquisition and socialization: Three develop- mental stories and their implications. In R. A. Shweder & R. A. LeVine (Eds.), Culture theory: Essays on mind, self, and emotion (pp. 276–320). New York: Cambridge University Press.Petitto, L. A., & Marentette, P. F. (1991). Babbling in the manual mode: Evidence for the ontog- eny of language. Science, 251, 1493–1496.Pinker, S. (1994). The language instinct: How the mind creates language. New York: Morrow.Shatz, M. (1987). Bootstrapping operations in child language. In K. E. Nelson & A. van Kleeck (Eds.), Children’s language: Vol. 6 (pp. 1–22). Hillsdale, NJ: Erlbaum.Shatz, M. (1992). A forward or backward step in the search for an adequate theory of language acquisition? Social Development, 1, 151– 154.Shatz, M. (1994). A toddler’s life. New York: Oxford University Press.Shatz, M. (in press). Revisiting A Toddler’s Life for The Toddler Years: Conversational participa- tion as a tool for learning across knowledge domains. In A. Brownell & C. B. Kopp (Eds.), Transitions in early socioemotional development: The toddler years. New York: Guilford.Shatz, M., & Ebeling, K. (1991). Patterns of language learning related behaviours: Evidence for self-help in acquiring language. Journal of Child Language, 18, 295–314.Slobin, D. (1985). The crosslinguistic study of language acquisition: Vols. 1 and 2. Hillsdale, NJ: Erlbaum.Slobin, D. (1992). The crosslinguistic study of language acquisition: Vol. 3. Hillsdale, NJ: Erlbaum.Slobin, D. (1997). The crosslinguistic study of language acquisition: Vols. 4 and 5. Hillsdale, NJ: Erlbaum.Tomasello, M. (1992). The social bases of language acquisition.[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Social Development, 1, 67–87.Tomasello, M. (2003). Constructing a language. Cambridge, MA: Harvard University Press.Werker, J., & Curtin, S. (2005). A developmental framework of infant speech processing. Language Learning and Development, 1, 197–234. 28. PART IBasic Foundations and TheoreticalApproaches to Language Development 29. IntroductionHow does the child achieve language? In this Part, five chapters describe five different,though not mutually exclusive, approaches to answering this question. Shafer andGarrido-Nag tackle the question at the physiological level. They begin with a primer onthe neurodevelopmental processes that underlie language development, and they reviewevidence of both similarities and differences between children and adults in how thebrain is organized for language functions. They describe the developmental changes thatoccur in the relation of function to underlying structure as sometimes resulting fromchanges in the type of processing the child engages in and sometimes resulting from thematuration of the structures themselves. They suggest neural underpinnings for critical-period phenomena and disorders of language development, and they describe a courseof development in which input shapes a genetically based initial structure. Goodluck answers the question of how children achieve language with the generativegrammar position that children have innate knowledge of Universal Grammar (UG).She then reviews the problem with this position, namely, that it must be reconciled withthe fact that children make errors. She proposes an account of the child language learneras having full access to UG but also as having processing limitations which accountfor the child’s failures to realize UG in actual comprehension and production ofsentences. Saffran and Thiessen review the evidence for the position that children achieve lan-guage by applying general learning procedures to the speech they hear. Although theprocesses they describe are not language specific, applying to both speech and nonspeechstimuli and used by both humans and nonlinguistic species, there are language-specificaspects to the application of these procedures. For example, infants selectively attend tospeech with the result that domain-general learning processes are particularly broughtto bear on language stimuli in infancy, and domain-general processes make use of lan- guage-specific knowledge throughout the course of development in guiding the analysesof input that yield further linguistic knowledge. 30. 20 Part I: Introduction Baldwin and Meyer argue that children’s social nature is an important part of theexplanation of children’s linguistic accomplishments. They bring clarity to this positionby making distinctions among three kinds of social factors relevant to language acquisi-tion: the social input children receive, children’s own social responsiveness, and children’ssocial understandings. They review evidence that usable language input is social (it mustcome from a person, not a screen), that children’s attentiveness to input is socially based,and that children’s social[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... understandings provide crucial support to the task of mappingsound to meaning. They present both the view and the controversies associated with theview that language is social and achieved through social means in each aspect: phonol-ogy, the lexicon, and syntax. Gathercole and Hoff argue that the way children achieve language is, to a significantdegree, by inducing structure from their analysis of input. They review evidence fromanalyses of the input available to the child, from computer simulations of acquisition,and from studies of the relation between variation in children’s input and children’sgrammatical development. They also conclude, however, that internal constraints mustoperate in the course of language development because input affects only the pace ofdevelopment, not the sequence in which the structures of language are acquired. Evans brings the theoretical framework of dynamic systems theory to bear on thequestion of how children achieve language. In doing so, she reformulates the question,asking how the self-organizing processes that are intrinsic to the child, in combinationwith the child’s past input, current state, and extrinsic dynamics of the immediatecontext, result in the emergence of language. In this view, language emerges in the childboth ontogenetically and at the moment of speech as a result of similar dynamic pro-cesses. The distinction between competence and performance and the distinctionbetween acquiring language and using language are discarded. In addition to presenting an array of current theoretical approaches to the field, thechapters in this Part illustrate the wide variety of methodologies that current researchemploys. The methods include the physiological technique of evoked potential recording(Shafer & Garrido-Nag), many different laboratory-based experimental techniques usedwith infants and children (Gathercole & Hoff; Goodluck; Saffran & Thiessen), com-puter simulations of language acquisition, and naturalistic, observational approaches(Gathercole & Hoff; Evans). The chapters in this Part also illustrate the broad range ofphenomena that characterize the field’s current definition of the accomplishment thatneeds to be explained. Some chapters focus on the acquisition of syntax (Goodluck;Gathercole & Hoff); others address phonological, lexical, and pragmatic developmentas well (Baldwin & Meyer; Evans; Saffran & Thiessen). Most focus on competence, butof necessity use performance as an indicator of competence. Evans argues that the dis-tinction is false. Together, the chapters in this Part on foundations and theoreticalapproaches provide a survey of how researchers in the field of language developmentcurrently frame the questions they ask and pursue answers to those questions. 31. 2The Neurodevelopmental Bases of LanguageValerie L. Shafer and Karen Garrido-NagImagine someone claiming to know how a car works on the basis of knowing how todrive, but having no notion of the internal mechanisms that make it work! If the carbreaks down, external signs (e.g., engine light, steam escaping from under the hood) areof little use in determining the cause, particularly for someone with no understandingof the internal mechanisms. It would also be virtually impossible to build a similar[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... oridentical system for powering a car on the basis of observing another car’s external per-formance because there are many different solutions that can lead to the same perfor-mance. Our understanding of language processing and language development whenbased entirely on behavior is limited in a similar manner. Knowledge of the neurobiology(i.e., internal mechanisms) is critical for arriving at a real understanding of languageprocessing and language development. The goal of this chapter is to illustrate how neurobiological evidence can help addressthree questions regarding language development. One issue concerns how biological andenvironmental factors interact during development to lead to the neurobiological spe-cialization for language found in the mature system. For example, why are certain lan- guage functions lateralized to the left and others to the right hemisphere? Directobservation of the brain mechanisms supporting linguistic processes and how theydevelop will help address this question. A second question is which processes contributing to language are limited by sensitiveor critical periods? Evidence from behavioral data suggests that there are sensitive periodsfor both first- and second-language learning. However, the complex nature of languageprocessing makes it difficult to isolate the specific causes of these limitations. Identifyingwhich brain structures/functions correspond to different language functions and whichof them have sensitive or critical periods for learning is necessary to answer thisquestion. Answers to both these questions are necessary to help further understand a thirdquestion: What are the biological and environmental causes of developmental language 32. 22 Valerie L. Shafer and Karen Garrido-Nagdisorders and delays? Current understanding of these disorders is primarily in terms ofdescriptions of impaired behaviors. Since a number of different causes could lead to thesame observed impairment, this level of knowledge is inadequate. Knowledge of theunderlying causes in terms of brain function should lead to earlier and more accurateidentification of a disorder and improved strategies for remediation. Returning to thecar analogy, it is easy to see that a mechanic will have greater success in repairing a carthat has overheated if the underlying cause is known. For example, a hole in the radiatorversus a broken fan can lead to the same breakdown, but each requires a differentrepair. The first section of this chapter provides a neurobiological model of the adult languagesystem. This model includes a description of structures involved in language processingand their corresponding functions as revealed by neurobiological studies. This sectionalso provides definitions of basic principles of neuroscience that are necessary for under-standing brain function and for interpreting neurophysiological data, and a descriptionof the event-related potentials (ERPs) method, which is most frequently used in neuro-developmental studies. The final sections focus on the three questions posed above. Our knowledge of neu-rodevelopment of language is rather limited at this time because, until recently, researchin this area has been sparse; even so, there is sufficient information to demonstrate howthese data can aid in addressing the questions of interest.Neurobiology of the Adult[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... SystemUntil recently, knowledge of neurobiology of language was gained from observing break-down in function due to brain damage. This method of study led to the view that majorfunctions of language were primarily localized to the left hemisphere, and that two largebrain regions, Broca’s area (roughly left inferior frontal cortex) and Wernicke’s area(roughly left posterior superior temporal cortex), were responsible for most languagefunctions. Damage to Broca’s area led to deficits in production, such as articulation,sequencing of speech sounds, and sentence production, whereas damage to Wernicke’sarea resulted in deficits of speech perception, and word and sentence comprehension.These observations led to the claim that Broca’s area was responsible for executing lan-guage and Wernicke’s area for speech perception and language comprehension, despitesome inconsistent evidence. For example, some Broca’s aphasics also showed deficits ofcomprehending certain syntactic structures (e.g., passives). Recent advances in neurobiological methods have dramatically increased our knowl-edge of brain structure–language function relationships in the mature adult system, andit has become clear that the broad two-system view is inaccurate. Functional brainimaging data from non-clinical populations indicate that the language system consistsof many interconnected small modules, each with a specialized task contributing tolanguage function (see Bookheimer, 2002). To understand language function in termsof the brain, the location and function of each of these supporting modules must beidentified, as well as the connectivity between modules. These relationships will be 33. The Neurodevelopmental Bases of Language 23briefly reviewed below, following a discussion of basic principles of neuroscience that arenecessary to understand how the brain works.Principles of neuroscienceThis section describes brain function in terms of neural–electrical circuits and how thesecircuits are used to store information and are modified in learning. This explanation ofbrain function will focus on the electrical function of neural circuits because informationstorage is in terms of firing of neurons, and much of our knowledge of neurodevelopmentcomes from examining electrical potential at the scalp using ERPs. The basis for storing and communicating information in the brain is via electro-chemical signals passed between neurons (cell bodies) via a long process called an axon.A chemical message (neurotransmitter) is released at a terminal (synapse) at the end ofan axon and received on another neuron, often at a dendrite (extension of the neuron).This chemical message leads to changes in the electrical potential of the receivingneuron. Sufficient change in potential (typically from receiving chemical messages frommultiple synapses) can lead to either excitation (and firing) or inhibition (which willmake firing more difficult). A neural circuit consists of axonal connections betweenthousands of neurons. Within a circuit, neurons fire in synchrony to particular stimuluscharacteristics (e.g., frequency) or processing demands (e.g., discriminating two stimuli).Different neural circuits can be linked together to accomplish more complex process-ing tasks.[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Language, like other functions, requires that information be learned and stored inmemory. The manner in which this learning occurs is often called Hebbian learning.Neurons become part of a neural circuit in two ways: First, there is some pre-wiring ofcircuits that is determined biologically, typically found at lower levels of the nervoussystem. Second, circuits are created by strengthening or weakening the synaptic connec-tions between neurons in the presence or absence of correlated firing patterns. Even inthe case of pre- wired circuits, stimulation is often needed to maintain the circuit, andfurther tuning of the circuit can occur based on input (see Hebb, 1949; Pulvermüller,1996; Vaughan & Kurtzberg, 1992). Memory for an association is the result of these synaptic modifications. Learning andmemory are intrinsic properties of the same neural networks, rather than being functionsof distinct brain structures. In other words, brain structures activated in learning a spe- cific stimulus or task are also activated in memory for the stimulus or task. Humans andother higher animals have evolved sophisticated systems to support learning and memoryof both linguistic and non-linguistic information. These include medial temporal lobestructures (limbic structure and hippocampus) involved in declarative/explicit memoryand basal ganglia/cerebellar structures involved in procedural/implicit memory (Squire& Kandel, 1999). In summary, memory for information is seen as synchronous electrical activity acrossneurons in a neural circuit responding to some stimulus attribute or task. Learning isthe result of modification (strengthening or weakening) of synaptic connections acrossneurons, leading to the creation of new neural circuits, or to the linking of a number of 34. 24 Valerie L. Shafer and Karen Garrido-Nagcircuits into higher-order circuits. The brain structures involved in learning informationare the same as those involved in memory for the information.Neuroanatomical and neurophysiological evidenceIdentifying the brain location and function of specialized modules that contribute todifferent behavioral functions can be achieved by examining the connectivity within andacross neural circuits in the brain (structure) and by examining their activation patterns(function). This section describes how structural and functional evidence are used toidentify specialized modules in the brain.Structural evidence. Structural evidence for specialized modules is seen as distinctivepatterns of connectivity and cell types (i.e., cytoarchitectonics) and has been used todefine Brodmann’s Areas (e.g., primary auditory cortex is BA 41). The organization ofneocortex into four to six layers of neurons, each reflecting a different type of connectiv-ity, has facilitated identification of these distinctive patterns. Primary sensory regions(e.g., visual, auditory) have a thick layer of neurons that receives information from theperipheral senses (Layer IV), whereas primary motor regions have a thick layer of neuronsthat sends information to the motor system (Layer V). By contrast, association regionshave a thick layer of neurons that sends and receives information to and from othercortical areas (Layers II and III). Layer I (closest to the scalp) consists primarily of den- drites of neurons in deeper layers and axons synapsing on these dendrites, or[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... passingthrough this layer (but with some other target), and Layer VI contains a heterogeneousmixture of neuron cell bodies and dendrites from cell bodies in Layers II and IV. Theneurons in these different layers also differ (e.g., pyramidal neurons are efferent projec-tion neurons and are larger and more abundant in Layer V than in Layers II and III)(Kandel, Schwartz, & Jessell, 2000; Kolb & Whishaw, 1996). In summary, these structural studies have identified brain regions (specifically,Brodmann’s Areas) that are likely to carry out different functions. However, the patternof connectivity into and out of these areas can only be used to make limited guessesabout function. Real- time functional measures are necessary to identify the precisefunction of brain regions, as will be illustrated in the next section.Functional evidence. Imaging methods, such as functional magnetic resonance imaging(fMRI) and positron emission tomography (PET), provide fairly precise information onwhich brain areas are activated in a language-processing task, but not when they areactivated. These methods indirectly index activation of neural circuits by measuringblood flow to discrete brain regions (Bookheimer, 2002). Both PET and fMRI havebeen extremely useful for identifying brain regions activated in adult language process-ing. However, neither method tolerates movement from participants, and PET is inva-sive; both factors limit their use with non- clinical child populations. The optimal method for studying typical neurodevelopment in infants and youngchildren is electrophysiology. In addition to its excellent temporal resolution, it is rela-tively tolerant of movement (unlike magnetoencephalography, MEG), and unthreaten- 35. The Neurodevelopmental Bases of Language 25ing to a parent or child. The development of multichannel systems for recordingthe electroencephalogram (EEG) has improved the spatial resolution of this method,although it remains poorer than fMRI, PET, and MEG. This section provides a briefdescription of the most commonly used electrophysiological method, event-relatedpotentials (ERPs). Electrophysiology takes advantage of the phenomenon that firing of neurons leads tochanges in the electrical potential (post-synaptic excitatory and inhibitory potentials) ofthe extracellular solution. These changes propagate to the surface of the scalp where theycan be amplified and measured using a voltage meter. Adjacent neurons in neural circuitsin many regions of the neocortex are arranged in parallel, so that the axons of the neuronsare aligned all on one end, and the inputs into dendrites on the other end. This arrange-ment leads to the circuit behaving like a dipole (positive at one end, negative at the other)when neurons in a neural circuit fire in synchrony. Thus, on one end (e.g., positive end),ions of the opposite polarity will be attracted, and this perturbation will propagate tothe scalp, while on the other end the opposite phenomenon will be observed. These tinyfluctuations from a set of circuits participating in one process (e.g., speech discrimina-tion) are too small to resolve in the unprocessed EEG, because it is the summation ofactivity from firing of many brain circuits. A number of signal-processing methods have been developed to isolate the electricalpatterns related to a single event of interest. The most commonly[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... used is that of averag-ing portions of the EEG time-locked to an event which has been repeated, with the goalof decreasing the contribution from processes that are not time-locked. Specifically, theactivity from events that are not time-locked is sometimes positive, and sometimes nega-tive, and will sum to zero as more trials are added. Time-locked events that are consis- tently of the same polarity and latency are maintained with averaging. For example, ininfants and children a positive polarity at fronto-central sites is found around 100 ms(P1 or P100) following the onset of an auditory event. This method is called averagedERPs. Figure 2.1 illustrates the instrumentation, unprocessed EEG, and an averagedERP to a vowel sound at multiple sites (fronto-central in gray and posterior-inferior inblack) that show opposite polarity for the child P100. This pattern of polarity (positiveon top of the head and negative at the back and below the ears) is consistent with aneural source in auditory cortex.1 Event-related potentials are useful for examining learning and memory because theycan provide information that participants can discriminate particular stimuli, with orwithout attention. They also have the power to reveal the sequence of processes leadingup to a behavioral response. Figure 2.2 shows auditory and language-dependent ERPcomponents to consonant–vowel–consonant (CVC) words in a match-to- sample phono-logical task. Participants were asked to decide whether the second word of a pair was thesame or different from the first word in the pair (e.g., cheese vs. cheev). The sequenceof early occurring peaks, N1 and P2 in adults, are evoked by the physical properties ofthe stimulus and are called obligatory components. Components which are modulatedby task or context and occur at later latencies are called cognitive or endogenous com-ponents. For example, the N400 component amplitude is modulated by a phonologicaldecision. Each component has a distinctive topography. For example, as shown in theright-hand graph of Figure 2.2, the endogenous phonological N400 is largest at lateral 36. 26 Valerie L. Shafer and Karen Garrido-Nag Ongoing EEG Amplifier S S S S 1s Auditory event-related potentials 8 Monolingual Infants 6 Standard 4 2 µV 0 –2 Signal averager –4 –6 –8 –100 0 100 200 300 400 500 600 700 ms Stimulus Stimulus onset P1Figure 2.1 Instrumentation and signal processing for recording ERP data. Brain activity isrecorded through electrodes (top left) and amplified, and the precise time of stimulus events isrecorded with the digitized EEG (top right). Multiple events are averaged so that the ERP emergesfrom the background noise (bottom right). Components indexing time-locked brain activity areseen as the time-points of greatest divergence (i.e., standard deviation) in amplitude across thesites, as seen at P1 in this graph plotting multiple sites.posterior sites (P7 and P8) and the N1 obligatory component is largest at superior frontalsites (FC1 and FC2). Possible neural sources of ERPs can be inferred from the scalp topography becauseof their dipolar nature. Electrodes placed on the scalp at opposite poles of the dipole willrecord opposite potentials. Source localization with ERPs is imprecise, but can beinformed by knowledge of neuroanatomy and information from other[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... imaging methodswith better spatial resolution (e.g., fMRI), and from clinical and animal research.Despite the limitations in source localization, different distributions of activity acrossthe scalp to two events, such as that seen in Figure 2.2, can be used to make inferencesthat different sources are activated in processing these events. In summary, the optimal method for studying neurodevelopment of language is event-related potentials because they have excellent temporal resolution and are suitable foruse with children. The latency and topography of ERP components can be used to make 37. The Neurodevelopmental Bases of Language 27 6 2 PZ (cheev)–cheev Topography (cheese)–cheev 1 4 cheev–(cheev) 0 2 –1 FC1µV µV 0 –2 FC2 P7 –2 P8 –3 –4 N400 –4 N1 –6 –5 0 200 400 600 800 ms 0 200 400 ms N1 P2 N400 P3/LPCFigure 2.2 Event-related potential components to CVC words in a match-to-sample task. Theleft-hand graph shows the obligatory N1 and P2 components and the cognitive components N400and P3/LPC. The thick and thin black lines show ERPs recorded to the second word in a pairand the dotted line to the first word in a pair. The right-hand graph shows subtraction of theERP to the second word in a different pair from that of the second word in a same pair andillustrates differences in topography for N1 and N400 (adapted from Shafer, Schwartz, & Kessler,2003). PZ, superior posterior midline site.inferences about the sequence of processes and cortical sources supporting languagefunctions.Brain indices of language processing in adultsOne goal of this chapter is to explain how neurodevelopment leads to the organizationand function of language observed in the adult brain. This section provides a descriptionof this endpoint of development, and is organized in terms of imaging and ERP studiesindexing (1) phonetic/phonological processing related specifically to speech perception,(2) morphosyntactic processing, focusing on anterior cortex involvement, and (3) seman-tic/discourse processing, focusing on comprehension. Figure 2.3 displays brain regionstypically activated in imaging studies.Phonetic/phonological processing. Imaging investigations of speech perception have iden-tified a number of brain structure–function relationships related to speech perception(see Bookheimer, 2002; Scott & Wise, 2004). These include primary auditory cortex(BA 41) and secondary auditory cortex (BA 42). Different portions of these regions areactivated in different types of processing. Activation in the bilateral superior temporalgyrus (STG) is seen to pre-lexical processing of phonetic features. The left posterior 38. 28 Valerie L. Shafer and Karen Garrido-Nag(a)(b)Figure 2.3 Brain structure–language function relationships. (a) Brain regions typically activatedduring (1) phonetic/phonological processing related specifically to speech perception, (2) mor-phosyntactic processing, and (3) semantic/discourse processing. (b) The premotor cortex (BA 6),pars opercularis, pars orbitalis, and pars triangularis (BA 44/45). 39. The Neurodevelopmental Bases of Language 29superior temporal sulcus (STS) shows particular sensitivity to phonetic features and theright STG to[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... melodic variation of speech. Mapping of speech onto lexical-semantic rep- resentations activates the left anterior portion of the STS (BA 22). The planum temporale(PT), which is in posterior superior temporal cortex, appears to serve as a motor/sensoryinterface for any acoustic stimulus. A few studies also suggest that left prefrontal cortex(BA 44/6) is activated in processing rapid transitions, such as those found in conso-nant–vowel syllables, and in accessing, sequencing, and monitoring phonemes. Thislatter function may be better characterized as working or phonological memory. Event-related potential evidence of activation of posterior cortex in phonetic/phono-logical processing comes largely from studies designed to elicit the ERP discriminativeresponse, mismatch negativity (MMN). The MMN is seen as a negative shift between100 and 300 ms at fronto-central scalp sites following the onset of a stimulus (called thedeviant) that is delivered among a series of repetitions of a more frequent sound (calledthe standard), and can be elicited without attention. Mismatch negativity reflects the phonological status of a pair of sounds in a listener’slanguage (e.g., Näätänen et al., 1997; Winkler et al., 1999). Specifically, MMN is earlierand/or larger to a pair of contrasts for speakers who have had experience with the soundsas phonemically contrastive. This difference in experience is typically observed over lefthemisphere sites (e.g., Näätänen, 2001; Shafer, Schwartz, & Kurtzberg, 2004). Dipolemodeling suggests that neural circuits in superior temporal cortex are a major contributorto MMN. In summary, the mature neurobiological system activated in phonological processingincludes posterior brain regions (BA 41, 42, 22) for speech perception, and PT forinterfacing with the motor system. These regions display somewhat different functionsfor left versus right hemisphere cortex. Activation of anterior regions in speech perceptionis probably related to phonological memory for sequential information. The ERP com-ponent MMN indicates that language-specific phonological categories affect processingat a pre-attentive level within 250 ms of a syllable onset. The MMN is not the only ERPcomponent sensitive to phonological variables, but tasks designed to elicit other compo-nents, such as the phonological N400 (see Figure 2.2), have not been widely used ininvestigations of young children.Morphosyntactic processing. Neurobiological studies indicate that anterior brain regionsplay a critical role in morphosyntactic processing. Some researchers suggest that thisanterior activation reflects structure-building processes, whereas others suggest that itreflects integrating syntax and semantics (see Bookheimer, 2002). Imaging studies show that processing of morphosyntactic information includes regionsin left prefrontal cortex, as shown in Figure 2.3. The premotor cortex (BA 6), parsopercularis, and putamen (basal ganglia structure) are activated in learning and recog-nizing simple grammars with phonological encoding, and pars opercularis and triangu-laris (BA 44/45) are activated in recognizing higher-level grammatical patterns. Inposterior cortical regions, the anterior portion of the left superior temporal gyrus is alsoactivated in processing morphosyntactic information (Friederici, 2004; Friederici,Ruschemeyer, Hahne,[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... & Fiebach, 2003). 40. 30 Valerie L. Shafer and Karen Garrido-Nag 1.9 1.0 Left anterior Neg 4 Word-order violation P600 0.5 0.9 3 0 2µV –0.5 0 µV 1 –1.0 0 –1.5 –0.9 –1 eLAN –2.0 LAN –2 0 200 400 600 Pos –1.9 0 200 400 600 800 ms msFigure 2.4 The left-hand graph displays ERPs at a left anterior site to a grammatical (thin line)compared with an ungrammatical sentence (*The zebra that the hippo kissed the camel on thenose] ran far away,” thick line). The topographical display (center) shows the negative (neg) andpositive (pos) distribution of the LAN (interpolated from all sites at 400 ms; adapted fromHestvik, Maxfield, Schwartz, & Shafer, in press). The right-hand graph shows the P600 to aword-order violation (adapted from Kessler, Martaharjono, & Shafer, 2004). Event-related potential indices of morphosyntactic processing also suggest a majorrole for anterior cortex. An early left anterior negativity (eLAN) occurring 100 to 300 msfollowing a violation is believed to index early stages of phrase-structure building, anda later left anterior negativity (LAN) reflects violations of morphosyntactic processes,such as use of gender information. A late positive component (P600), which can co-occurwith eLAN or LAN, is found to violations of syntax and less preferred morphosyntacticstructures and indexes integration of syntactic and semantic information (Friederici,2002). Figure 2.4 displays the eLAN, LAN, and P600 to morphosyntactic violations.Studies comparing function words and content words have also shown a larger early leftanterior negativity to function words than to content words (Neville, Mills, & Lawson,1992; Pulvermüller, 1996). By contrast, sustained anterior positivities (SAPs) are seen to function words at theonset of sentences in discourse, and continue as long as the words in the utterances aremeaningful, as shown in Figure 2.5 (Shafer, Kessler, Morr, Schwartz, & Kurtzberg,2005). Shafer and colleagues (2005) suggested that the left SAP reflects automatic build-ing of grammatical structure and the right reflects building discourse structure becauseonly the right drops out when attention is directed away from the auditory story to thestory-line of a silent movie. The anterior scalp locations of eLAN, LAN, and SAP are consistent with sources ininferior frontal cortex or in the anterior portion of superior temporal cortex. Anteriorcortex and basal ganglia structures (involved in implicit learning of patterns) appear toplay a role in the generation of P600 since damage to left inferior frontal regions (Broca’saphasia) but not to left temporo-parietal regions leads to reduced and delayed P600(Friederici, Kotz, Werheid, Hein, & von Cramon, 2003; Wassenaar & Hagoort, 2005)and damage to basal ganglia structures reduces or abolishes the P600, but not eLANand LAN components (Friederici, Kotz, et al., 2003). In summary, both imaging and neurophysiological studies indicate that left anteriorregions, including Broca’s area and basal ganglia subcortical structures, are highly acti-vated in early and late aspects of morphosyntactic processing. 41. The Neurodevelopmental Bases of Language 31Semantic/discourse[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... processing. Research from clinical studies led to the view that poste-rior brain regions, particularly Wernicke’s area, are the major contributors to semanticand discourse processing. Recent studies of non-clinical adults, however, have shownthat anterior as well as posterior regions carry out important semantic and discoursefunctions. This section describes the principal regions and timing of activation of theseregions in semantic/discourse processing. Imaging studies have revealed that widespread anterior and posterior regions are acti-vated in semantic processing, and that this activation is systematic and predictable interms of semantic and sensorimotor properties (Bookheimer, 2002). For example, image-able words (lexemes), such as animals and concrete concepts, activate visual cortex(occipital lobe) and association cortex involved in object identification (e.g., inferior tem-poral gyrus), whereas graspable lexemes also activate anterior motor and premotor cortex(e.g., BA 6) associated with reaching and grasping. The pars orbitalis of the left inferiorfrontal gyrus is also activated in processing semantic relationships and/or retrievingsemantic information. As mentioned above, the left superior temporal sulcus (posteriorcortex) is activated in mapping between speech and lexical- semantic representations. Right cortical regions, typically homologous to the left language regions, are activatedin pragmatic/discourse and prosodic functions. These functions include interpretingmetaphors and morals, creating coherence, topic maintenance, and using prosody tointerpret emotion. For example, right regions, including BA 44/45, dorsolateral prefron-tal cortex (BA 46), superior temporal cortex (BA 22), and angular gyrus (BA 39) arehighly activated in topic maintenance (Caplan & Dapretto, 2001). Bookheimer (2002)suggested that the right hemisphere activation reflects integration of information overtime, whereas left activation during language comprehension indexes interpreting themeaning of individual units. Event-related potential studies of semantic/discourse processing typically show theeffects of semantic processing beginning around 200 ms following a word form. Themost commonly examined index of semantic processing is the N400, which peaksbetween 200 and 500 ms following the word of interest (e.g., Kutas & Hillyard, 1980).The increased N400 to unprimed compared with primed words in priming studies isprobably due to activation of phonological and semantic neighbors in the process oflexical access (e.g., Praamstra, Meyer, & Levelt, 1994; Shafer et al., 2003, see Figure2.2). In sentence- processing studies, the N400 negativity additionally reflects integrationof a word into sentence context. Several studies have observed differences in N400topography to semantic categories (e.g., motor vs. imageable) consistent with fMRIfindings (Pulvermüller, 1996). Dipole analysis of the N400 identifies sources in left prefrontal cortex, beginningaround 250 ms after word onset, followed by activation of sources in left and right lateralprefrontal cortex and right inferior temporal cortex from 350 to 450 ms (Frishkoff,Tucker, Davey, & Scherg, 2004). The activation of frontal sources was sustained (250to 800 ms), similar to the SAP seen by Shafer and colleagues (2005, Figure 2.5).[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Thesedata support the view that the left frontal region is involved in executive control oversemantic processing and the right over discourse processing. In summary, multiple cortical regions are activated in semantic and discourse pro-cessing. The pars orbitalis portion of the inferior frontal gyrus probably functions in 42. 32 Valerie L. Shafer and Karen Garrido-Nag 6 FC5 Story Nonsense SAP 4 2µV 0 –2 0 200 400 600 800 msFigure 2.5 The graph shows the sustained anterior positivity (SAP) of the ERP to “the” followedby story context (e.g., “the curious little kitten, etc.”) compared with that of “the” followed bynonsense (e.g., “the gikopo, etc.”). The head image shows the anterior topography of this positiv-ity (peak at the center of concentric circles; adapted from Shafer, Kessler, et al., 2005).executive control, and activation is sustained throughout processing. Activation of otherregions reflects retrieval of the semantic properties of a word. Right hemisphere activationis specifically involved in pragmatic/discourse aspects of processing.General Principles of Developmental NeuroscienceSeveral important principles of neurodevelopment are described in this section, beforewe examine neurodevelopment of language. These principles lead to predictions aboutthe timecourse of development of different types of information processing and arenecessary to help interpret neurodevelopmental ERP data. The timing of development of different brain regions typically proceeds fromlower to higher levels. This timecourse of maturation places limitations on when func-tions can emerge, and later emergence is expected for higher-level functions. Specifically,neurons and axon projections mature earlier in peripheral systems (e.g., brainstem)compared with central systems (e.g., neocortex); within neocortex, primary motor andsensory regions mature earlier than secondary regions, which in turn mature earlierthan association regions (e.g., Kolb & Whishaw, 1996). This progression of develop-ment has been determined largely through examining synaptogenesis, myelination, andneuron death. Synaptogenesis is creation of synaptic sites on which axons can connect. Subcorticaland primary cortical areas are relatively mature in newborns and reach a peak in synap-togenesis by 3 months of age. Association cortex, such as prefrontal regions, does not 43. The Neurodevelopmental Bases of Language 33reach peak synaptogenesis until around 3 years of age. These peaks are followed byloss of synaptic sites and loss of neurons related to absent or weak connections (e.g.,Huttenlocher & Dabholkar, 1997). Myelination is the creation of a myelin sheath on axons, which leads to fastertransmission of signals. Primary regions complete myelination quite early, beginningbefore birth, followed by secondary auditory cortex, and lastly by association cortex.Some regions do not complete myelination until well past puberty (Vaughan &Kurtzberg, 1992). Brain regions can function before reaching peaks in synaptogenesis and beforemyelination is complete. However, a peak in number of synapses suggests that the brainregion is ripe for learning, since many sites for axonal[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... connections are available. Comple-tion of myelination within a neural circuit indicates that rapid processing can take place.Infants and children typically show later latencies than adults for a number of ERPcomponents, in part reflecting incomplete myelinations (e.g., N1: Ponton, Eggermont,Kwong, & Don, 2000; MMN: Shafer, Morr, Kreuzer, & Kurtzberg, 2000; N400: Coch,Maron, Wolf, & Holcomb, 2002; Friedrich & Friederici, 2004; eLAN, LAN: Hahne,Eckstein, & Friederici, 2004). They also show larger-amplitude ERP components, prob-ably reflecting a greater abundance of neurons and synapses, less specificity in firing, andpoorer insulation of the axons by myelin. Finally, some adult ERP components may beabsent or attenuated in children’s responses and others enlarged, particularly to stimulipresented at fast rates. For example, the early obligatory component P1, generated inLayer I cortex, dominates children’s ERPs, whereas N1, generated in Layers II and III,is absent at fast rates (compare Figures 2.1 and 2.2). This pattern is related to earlymaturation of Layer I compared with Layers II and III (Ponton et al., 2000). These maturational changes in brain neurobiology will affect processing of speechand language stimuli. One challenge in developmental neurobiology is to tease apartwhich changes are due to general maturation and which are related to learning a specificlanguage; in other words, what aspects of development are related to environmental inputversus genetic instruction. Higher levels of processing are increasingly dependent oninput (Greenough, Black, & Wallace, 1987; Knudsen, 2004). For example, in someneural circuits, a certain type of information is expected (experience-expectant) andnecessary to maintain the circuit and “tune” it. Other circuits (experience-dependent),generally in association cortex, need input, but do not require a certain type of informa-tion. There are also lower- level brain regions that are hard-wired to process a certaintype of information (e.g., hard-wired connections from auditory periphery). However,even in these cases, sensory or motor activation may be necessary to maintain thecircuits. The location of primary and secondary cortical regions activated in processing thesensory and motor aspects of speech is largely determined genetically. The developmentalbasis of the organization of higher-level processing, however, is less clear. The locationof experience-dependent circuits appears to be related to the nature of the stored infor-mation. For example, association cortex near visual regions often stores higher-level rep- resentations that include visual information and are connected to circuits in proximalprimary and secondary visual cortex. However, association cortex can also be con-nected to distal regions (e.g., prefrontal cortex to occipital regions). Currently there are 44. 34 Valerie L. Shafer and Karen Garrido-Naginsufficient data describing the initial state (i.e., in newborns) of connectivity within andacross association cortex. A number of studies have also found sex differences in neurodevelopment, seen ashemispheric differences in processing speech and language, particularly for infants under9 months of age (e.g., Shafer, Shucard, & Jaeger, 1999; Shucard & Shucard, 1990). Thesesex differences are[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... another source of variability in studies of infants that can confoundinterpretation of hemispheric specialization for language. In summary, lower levels of the nervous system mature earlier and are less dependenton sensory or motor input/activation than higher levels. Based on these observations,developmental studies of language should show less change across age in processing atlower levels and more change at higher levels, related to language input.Neurodevelopment of Speech and LanguageAs revealed above, research over the past two decades has greatly increased our knowl-edge of neurobiology of language in the mature system. However, current knowledge ofneurodevelopment is quite limited because the necessary studies are only beginning tobe done. Even so, there are sufficient data to begin to address the three questions posedin the introduction, as will be illustrated below.What is the developmental basis of adult language organization in the brain?The first question of interest is how biological and environmental factors interact duringdevelopment to lead to the neurobiological specialization for language found in the adultsystem. To address this question, we will focus on brain organization in semantic pro- cessing and morphosyntactic processing.Semantics. The description of organization of adult semantic processing revealed thatactivation of brain regions in processing a certain stimulus or in performing a semantictask was not random. For example, processing words that can be grasped activatedanterior motor and premotor regions that are used in the motor activity of grasping.How does this organization develop? Research by Mills and colleagues shows developmental changes in ERP activity, ini-tially seen as broadly distributed negativity (over left and right hemispheres) to knownwords in infants with smaller vocabularies or to newly learned words (Mills, Prat, &Zangl, 2004). This negativity may reflect lexical-semantic processing indexed by the adultN400. This activity becomes left-lateralized with increasing vocabulary size. Mills andcolleagues suggest that the developmental change in topography reflects rate of acquisitionand number of exposures to a word (Mills & Conboy, 2005). The more focal left hemi-sphere activity may be related to increased specificity in neural firing. That is, synapsesbetween neurons that do not fire in synchrony to a word will be weakened and lost,leaving a smaller, more focal population of neurons that is responsive to the word. 45. The Neurodevelopmental Bases of Language 35 The claim that regions activated in learning are also activated in memory suggests afurther interpretation of the data. Evidence from adult neurobiology indicates that righthemisphere activation is related to discourse aspects of processing, which include episodicmemory, that is, memory for the situation and language context in which the word wasused. Research has shown that first encounters with words lead to partial knowledge,which is typically context-bound (e.g., Wagovich & Newhoff, 2004). Thus, the righthemisphere activation seen in infants to newly learned words may reflect the same dis-course-related functions as observed in adults. Increasing exposure to a word will lead to[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... more context-free knowledge, and lessactivation of right hemisphere structures for a word presented out of discourse. Networksactivated in context-free representations of word meaning probably favor the left hemi- sphere as an indirect consequence of preferential processing of the phonetic features inthe left hemisphere. Specifically, if more neural circuits for processing phonetic featuresare in the left superior temporal sulcus, then a greater number of connections can bemade between semantic-representation circuits and phonetic-form circuits in this hemi-sphere. Dehaene-Lambertz and Gigla (2004) found evidence of greater left than righthemisphere activation in both speech and non-speech discrimination in infants, andsuggested that left hemisphere lateralization for language in the mature system may arisefrom this developmental bias in auditory processing. Under this interpretation, the func-tion/structure relationships found in adults are already present in toddlers. In summary, these studies illustrate how neurodevelopmental data can be used tobegin addressing the neurodevelopmental basis of semantic representation. These dataare consistent with the claim that the structures involved in learning a particular typeof information (in infancy or adulthood) are the same as those involved in memory forthat information. Specifically, the pattern of activation of the right and left hemisphereto words in toddlers can be explained in terms of the adult system. Future studies willneed to examine the neurodevelopment of different semantic categories (e.g., tools vs.animals) to further substantiate this claim.Morphosyntax. Brain structural studies indicate that the timecourse of maturation ofanterior regions BA 44/45 is slow. Specifically, BA 45 (premotor cortex) does not reachadult structural asymmetry until age 5, and BA 44 (Broca’s area) not until age 11(Amunts, Schleicher, Ditterich, & Zilles, 2003). Behavioral studies also show subtledifferences in processing syntax up to age 11 (see Hahne et al., 2004, for discussion). The few studies that have examined ERPs to morphosyntax in young children alsosuggest slow maturation of function in these regions. Young children do not exhibit thetypical eLAN and LAN observed in adults. Specifically, early positivities rather thannegativities are elicited to violations in 2-year-old (Oberecker, Friedrich, & Friederici,2005) and 6-year-old children (Hahne et al., 2004). In our laboratory, we also haveobserved this early positivity to structural violations in 4-year-old children, as shown inFigure 2.6. Following the early positivity, LAN and P600 are seen to violations in simpleactive sentences for 2-year-olds (31–34 months of age: Oberecker et al., 2005), and inpassive sentences for 7- to 10-year-old children (Hahne et al., 2004). The early positivityis no longer present in these older children. The SAPs observed to the functionword “the” in stories are also more posterior for children than adults, perhaps reflecting 46. 36 Valerie L. Shafer and Karen Garrido-Nag 15 Adult anterior sites “the” left “the” right 10 “ko” left “ko” right 5 µV 0 Early NEG NEG –5 15 Child anterior sites POS 10 5 µV 0 –5 NEG 0 200 400 600 800 msFigure 2.6 Four-year-old children show an early positivity and later negativity, unlike[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... adults, tothe nonsense syllable (ko) replacing the function word “the” in stories.immaturity of anterior cortex (cf. Shafer, Schwartz, Morr, Kessler, & Kurtzberg, 2000,and Shafer, Kessler, et al., 2005). In summary, the finding of LAN and P600 to structural violations in 2-year-oldchildren suggests that brain structures activated in adult processing are sufficientlymature by the third year of life to support morphosyntactic learning. Differences betweenchildren and adults are found, particularly in the presence of an early positivity for thechildren under 7 years of age, and later latencies of the LAN and P600. The functionalsignificance of the positive component to deviant patterns needs to be determined toallow further explanation of these developmental differences, although Oberecker et al.(2005) suggest that it reflects “heightened” sensitivity to acoustic cues. These datasupport the claim that structures involved in language processing in the adult areinvolved in learning in the child.Which processes contributing to language are limited bysensitive/critical periods?The second question posed in the introduction was which processes contributing tolanguage are limited by sensitive/critical periods? There is reasonably good evidence from 47. The Neurodevelopmental Bases of Language 37 2 Monolingual 2 Early bilingual 2 Late bilingual FZ Standard FZ Deviant 1 1 1 FZ Difference 0 0 0µV µV µV –1 –1 –1 MMN MMN MMN? –2 –2 –2 0 200 400 600 0 200 400 600 0 200 400 600 ms ms msFigure 2.7 Mismatch negativities (MMNs) to the [I]/[ε] contrast for monolingual speakers ofEnglish (left), early bilinguals (center), and later learners of English with Spanish as a secondlanguage (from Garrido, Hisagi, & Shafer, 2005).behavioral studies of late learners of a first language (L1) or second language (L2) indi-cating that there are sensitive periods for learning native speech categories and morpho-syntactic patterns (Johnson & Newport, 1991; Strange & Shafer, in press). However,the complex nature of language processing makes it difficult to determine what underly-ing processes lead to these limitations. Neurophysiological data can inform this questionby revealing which underlying processes and corresponding brain regions are the sourcesof these limitations. Developmental data are needed to reveal how age of exposure andamount of exposure interact in development of the brain circuits subserving languageprocessing. Event-related potential studies of adult L2 learners from our laboratory suggest thatthere may be an early sensitive period for learning of language-specific speech categories,which results in automatic selection of relevant speech cues. Specifically, early Spanish–English bilinguals with Spanish as L1 and who have learned English before age 5 havesmaller- amplitude mismatch negativities than monolingual English listeners to theEnglish phonemic contrast /I/ versus /ε/, as shown in Figure 2.7. This result contrastswith native-like identification and discrimination of these vowels in behavioral tasks bythese early L2 learners. As expected, the MMN is absent for first-language (L1) Spanishspeakers who know little English, because they have had little or no experience withthese phonemes until late in life. These results suggest that early L2 learners need atten-tion for native-[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... like performance, and thus show less automaticity in processing. However,the larger MMNs to L2 contrasts for proficient compared with novice L2 users in thisas well as other studies indicate that some improvement in automaticity of processingthese L2 categories can occur with L2 experience (e.g., Winkler et al., 1999; Winkler,Kujala, Alku, & Naatanen, 2003). A few ERP investigations of morphosyntactic processing in adults also suggest anearly sensitive period for learning morphosyntactic patterns. Specifically, only early L2learners of English showed an eLAN to morphosyntactic violations (Kessler et al., 2004)or left anterior negativity to function words (Weber-Fox & Neville, 1995). Note, however,that later-latency LANs may be a function of delayed phonological processing leadingto delayed access to grammatical information. These studies suggest that automatic, native-like processing requires learning ofspeech contrasts and morphosyntactic patterns in a narrow time-window, perhaps during 48. 38 Valerie L. Shafer and Karen Garrido-Nag Left frontal monolingual Midline frontal monolingual 8 8 Standard 6 Pos MMR 6 Deviant 4 4 DifferenceµV µV 2 2 0 0 –2 –2 Neg MMR Late Neg 0 200 400 600 0 200 400 600 ms ms 8 Left frontal bilingual 8 Midline frontal bilingual 6 6 4 Pos MMR 4µV µV 2 2 0 0 –2 –2 Neg MMR Late Neg 0 200 400 600 0 200 400 600 ms msFigure 2.8 Event-related potentials to 250 ms [I]/[ε] contrast in 20- to 36-month-old monolin-gual and bilingual infants at left frontal (F3) and midline frontal (FZ) sites.the first two or three years of life. However, data from children during this early periodof development are necessary to determine how input changes processing in these cir-cuits. Very little developmental data are available addressing this question and there aresome uncertainties regarding the interpretation of infant ERP components. Even so, anumber of recent studies are beginning to resolve the uncertainties in interpretation(e.g., Kushnerenko, Ceponiene, Balan, Fellman, & Naatanen, 2002; Morr, Shafer,Kreuzer, & Kurtzberg, 2002; Weber, Hahne, Friedrich, & Friederici, 2004) and twostudies suggest that an MMN-like negativity can serve as an index of the developmentof speech categories, at least at 12 months of age (Cheour, Alho, & Ceponiene, 1998;Rivera-Gaxiola, Klarman, Garcia-Sierra, & Kuhl, 2005). Specifically, these two studiesfound a larger ERP negativity to native than non-native contrasts in 11- and 12-month-old infants. The results for 6-month-old infants in these studies are less easily interpretedbecause it appears that some infants showed positivities and others showed negativities.Preliminary data from our laboratory show smaller positive mismatch responses (MMRs)and later and smaller negative MMRs to an [I]/[ε] contrast in toddlers with bilingualexposure to English and Spanish compared with those with monolingual English expo-sure, as shown in Figure 2.8. However, we currently have data from twice as manybilingual toddlers, and will need more monolingual participants to determine whetherthis apparent difference is significant. In summary, these studies illustrate that neurophysiological data can provide usefulinformation regarding the sensitive periods for language development. Future studiesneed to examine how age[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... of acquisition and amount of exposure affect the developmentof automaticity of speech processing. The infant ERP discriminative components also 49. The Neurodevelopmental Bases of Language 39need further research to establish their functional significance and developmentaltimecourse.What are the causes of developmental language disorders and delays?Understanding of developmental language disorders is limited without knowledge of theinternal mechanisms supporting language function, knowledge of which mechanismsare dysfunctional, and knowledge of how these mechanisms develop in typical andatypical cases. Researchers are beginning to identify neurobiological correlates of differ-ent types of developmental language disorder, such as specific language impairment(SLI), and a few studies have begun to examine the development of these processes inat-risk infant populations (e.g., Friedrich, Weber, & Friederici, 2004; Leppaanen, Pihko,Eklund, & Lyytinen, 1999; Molfese, 2000). This section will focus on SLI to illustratehow knowledge of typical versus atypical brain function and structure in processinglanguage can help elucidate this disorder. Specific language impairment is a developmental disorder specific to language in thatlanguage ability is significantly depressed compared with typical age- matched children,but non-language cognitive skills are within the normal range (see Rice, this volume).These children have particular difficulty with phonology and morphosyntax, and somemay also have poor auditory processing. The causal nature of this disorder continues tobe debated. As will be shown below, neurobiological data can aid in identifying possiblecauses.Deficits in speech perception. Event-related potential research on typical populationssuggests that by 12 months of age the speech-sound memory representations that areaccessed in the discriminative process indexed by MMN are weighted to reflect relevantcues of the ambient language. Studies of children with SLI have found absent and/orlate MMNs to phonetically similar speech contrasts (Shafer, Morr, Datta, Kurtzberg, &Schwartz, 2005; Uwer, Albrecht, & von Suchodoletz, 2002). These findings, along withpoor categorical perception in many of the SLI children, led to the suggestion that somechildren with SLI do not have correctly weighted phonological representations in audi-tory cortex. By contrast, the nature of the deficit for the children with SLI who showedlate MMNs and good categorical perception may be one of less automatic processing(Shafer, Morr, et al., 2005). Event-related potential evidence also supports the suggestion that poor phonologicalprocessing, for at least some children with SLI, is the consequence of poor auditoryprocessing abilities. One study found immature obligatory components (P1, N1, P2complex) in adolescent and teenage children with SLI (Bishop & McArthur, 2005). Inanother study, 6-month-old infants at risk for SLI were shown to have a reduced positiveMMR to a pitch change in tone pairs with a short interstimulus interval (ISI) of 70 ms(Choudhury, Friedman, Realpe-Bonilla, Chojnowski, & Benasich, 2005). In summary, these studies suggest that deficient processing of sounds in auditorycortex contributes to the types of language impairments found in[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... some children withSLI. This knowledge will help determine the causes of SLI and remediate this disorder 50. 40 Valerie L. Shafer and Karen Garrido-Nagbecause it will be possible to explore how different factors, such as attention and training,affect processing at this level.Morphosyntax. Deficient learning of morphosyntax in SLI could be the consequenceof a number of different causes, such as less salient phonetic substance of function wordsand inflections, limited cognitive resources (e.g., working memory) for learning complexpatterns such as morphosyntax, or a specific deficit in learning hierarchical patterns (i.e.,phrase structure). Knowledge of the brain structures and functions supporting process-ing of morphosyntactic information in typical and atypical development will help selectamong these various causes. The few studies examining morphosyntactic processing in children with languageimpairment (some of whom are SLI) suggest deficient processing in left perisylviancortex. Neville and colleagues found bilateral rather than the typical pattern of left- greater-than-right anterior negativity to function words in reading sentences in grade-school children with language learning impairments (Neville, Coffey, Holcomb, &Tallal, 1993). Shafer and colleagues observed reduced positivity to the function word“the” in stories or nonsense contexts at left temporal sites and increased positivity at righttemporal sites in grade-school children with SLI compared with children with typicallanguage (Shafer, Schwartz, et al., 2000). The increased right hemisphere positivity mayreflect greater use of semantic/discourse processing to compensate for poor left hemi-sphere function. The neural sources of these components may be left inferior frontal gyrus or left lateraltemporal cortex. Evidence from structural neuroimaging studies (MRI) is consistentwith these as possible sources. Specifically, adolescent children with SLI do not show thetypical left-greater-than-right structural asymmetry in the planum temporale, pars tri-angularis, and premotor cortex (e.g., Gauger, Lombardino, & Leonard, 1997; Jernigan,Hesselink, Sowell, & Tallal, 1991; Plante, Swisher, Vance, & Rapcsak, 1991). In summary, neurobiological studies suggest that deficits in left hemisphere processingunderlie poor morphosyntactic processing. Although these studies do not definitivelydemonstrate the neural source of the deficit, they provide sufficient evidence to narrowthe possible candidate structures. It is quite likely that neurobiological studies will showthat there are several different patterns of deviant brain processing. For example, somechildren may show atypical processing in posterior auditory cortical regions and othersin anterior cortex. These different patterns would suggest different causes contributingto SLI.ConclusionThe goal of this chapter was to show how knowledge of the neurodevelopmental bases oflanguage can advance our understanding of language organization in the mature system,and of the causes of developmental disorders of language. This area of study is relativelynew, and there are considerable gaps in our knowledge. Even so, evidence from the avail-able studies provides a starting point for examining these questions. These studies suggest[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... 51. The Neurodevelopmental Bases of Language 41that the precursors of adult brain organization can be found in infants. For example, righthemisphere processing of newly learned words in toddlers is consistent with the discoursefunction of right hemisphere structures in adults, and left hemisphere dysfunction inchildren with SLI suggests that this hemisphere plays a special role in language acquisition,as well as adult language functions. Neurobiological studies also are beginning to revealwhich brain structures/functions have sensitive or critical periods. For example, failure toset up phonological representations in left auditory cortex in the first few years of lifeappears to limit automaticity of processing this information. Finally, neurobiologicalstudies are helping to identify atypical brain function in developmental language disor-ders. In particular, atypical processing in left hemisphere structures were shown to con-tribute to deficits of phonological and morphosyntactic processing in children with SLI. Future research needs to carefully trace the development of components found inadults back to their initial emergence during the first few years of life to help clarify thefunctional significance of infant ERP components. These studies are necessary to dem- onstrate that ERP indices of language processes seen in adults indeed have precursorsearly in development. The recent improvements in instrumentation and data analysistechniques should allow research on the neurodevelopment of language to flourish inthe coming years. It is hoped that the increased ease of doing this research with infantsand toddlers will encourage more laboratories to pursue these questions.NotesWe thank Monica Wagner for providing thoughtful comments on the chapter. We are alsoindebted to the editors, Erika Hoff and Marilyn Shatz, for suggestions on how to approach thistopic, which have made this chapter much more accessible to the prospective reader. The writingof this chapter was supported, in part, by NIH HD046193.1 Electrical activity at a given site is always recorded with respect to activity at another active site called the reference. Thus the morphology of the waveform is dependent on activity at both sites (e.g., FZ relative to mastoid reference). However, the difference in amplitude between any pair of sites will be identical regardless of reference. For example, (FZ to mastoid ref.) − (CZ to mastoid ref.) = (FZ − CZ) + (mastoid ref. − mastoid ref.) = FZ − CZ, where FZ is fronto-central midline and CZ is central midline.ReferencesAmunts, K., Schleicher, A., Ditterich, A., & Zilles, K. (2003). Broca’s region: Cytoarchitectonic asymmetry and developmental changes. Journal of Comparative Neurology, 465, 72–89.Bishop, D. V. M., & McArthur, G. M. (2005). Individual differences in auditory processing in specific language impairment: A follow-up study using event-related potentials and behavioural thresholds. Cortex, 41, 327–341.Bookheimer, S. (2002). Functional MRI of language: New approaches to understanding the cortical organization of semantic priming. Annual Review of Neuroscience, 22, 151– 188.Caplan, R., & Dapretto, M. (2001). Making sense during conversation: An fMRI study. Neuro- Report, 12, 3625–3632.[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... 52. 42 Valerie L. Shafer and Karen Garrido-NagCheour, M., Alho, K., & Ceponiene, R. (1998). Maturation of mismatch negativity in infants. International Journal of Psychophysiology, Special Issue: Event related potentials and information processing by infants, 29, 217–226.Choudhury, N., Friedman, J. T., Realpe-Bonilla, T., Chojnowski, C., & Benasich, A. (2005, April). Difference in cortical responses to auditory stimuli in six-month-old infants with family history of specific language impairment and age-matched controls. Poster session presented at the 12th Annual Meeting of the Cognitive Neuroscience Society, New York.Coch, D., Maron, L., Wolf, M., & Holcomb, P. J. (2002). Word and picture processing in chil- dren: an event- related potential study. Developmental Neuropsychology, 22, 373– 406.Dehaene-Lambertz, G., & Gigla, T. (2004). Common neural basis for phoneme processing in infants and adults. Journal of Cognitive Neuroscience, 16, 1375–1387.Friederici, A. D. (2002). Towards a neural basis of auditory sentence processing. Trends in Cognitive Sciences, 6, 78–84.Friederici, A. D. (2004). Processing local transitions versus long-distance syntactic hierarchies. Trends in Cognitive Sciences, 8, 245–247.Friederici, A. D., Kotz, S. A., Werheid, K., Hein, G., & von Cramon, D. (2003). Syntactic comprehension in Parkinson’s disease: investigating early automatic and late integrational processes using event-related brain potentials. Neuropsychology, 17, 133–142.Friederici, A. D., Ruschemeyer, S. A., Hahne, A., & Fiebach, C. J. (2003). The role of left inferior frontal and superior temporal cortex in sentence comprehension: Localizing syntactic and semantic processes. Cerebral Cortex, 13, 170–177.Friedrich, M., & Friederici, A. D. (2004). N400 like semantic incongruity effect in 19-month olds: processing known words in picture contexts. Journal of Cognitive Neuroscience, 16, 1465–1477.Friedrich, M., Weber, C., & Friederici, A. D. (2004). Electrophysiological evidence for delayed mismatch response in infants at risk for specific language impairment. Psychophysiology, 41, 772–782.Frishkoff, G. A., Tucker, D. M., Davey, C., & Scherg, M. (2004). Frontal and posterior sources of event- related potentials in semantic comprehension. Cognitive Brain Research, 20, 329–354.Garrido, K., Hisagi, M., & Shafer, V. L. (2005, April). ERP indices of speech processing in Spanish–English bilinguals. Poster presented at the 12th Annual Meeting of the Cognitive Neuroscience Society, New York.Gauger, L., Lombardino, L. J., & Leonard, C. M. (1997). Brain morphology in children with specific language impairment. Journal of Speech Language and Hearing Research, 40, 1272–1284.Greenough, W., Black, J., & Wallace, C. (1987). Experience and brain development. Child Development, 58, 539–559.Hahne, A., Eckstein, K., & Friederici, A. D. (2004). Brain signatures of syntactic and semantic processes during children’s language development. Journal of Cognitive Neuroscience, 16, 1302–1318.Hebb, D. (1949). The organization of behavior: A neurophysiological theory. New York: Wiley.Hestvik, A., Maxfield, N., Schwartz, R. G., & Shafer, V. L. (in press). Brain responses to filled gaps. Brain and Language.Huttenlocher, P. R., & Dabholkar, A. S. (1997). Regional differences in synaptogenesis in human[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... cerebral cortex. Journal of Comparative Neurology, 387, 167–178.Jernigan, T. L., Hesselink, J. R., Sowell, E., & Tallal, P. A. (1991). Cerebral structure on magnetic resonance imaging in language and learning-impaired children. Archives of Neurology, 48, 539–545. 53. The Neurodevelopmental Bases of Language 43Johnson, J. S., & Newport, E. L. (1991). Critical period effects on universal properties of language: the status of subjacency in the acquisition of a second language. Cognition, 39, 215–258.Kandel, E., Schwartz, J., & Jessell, T. (2000). Principles of neural science. New York: William Heinemann and Harvard University Press.Kessler, K. L., Martaharjono, G., & Shafer, V. L. (2004). ERP evidence of grammatical process- ing in Spanish second language learners of English. In Proceedings of the 28th Annual Boston University Conference on Language Development (pp. 294–305). Somerville, MA: Cascadilla Press.Knudsen, E. I. (2004). Sensitive periods in the development of the brain and behavior. Journal of Cognitive Neuroscience, 16, 1412–1425.Kolb, B., & Whishaw, I. Q. (1996). Fundamentals of human neuropsychology. USA: W.H. Freeman and Company.Kushnerenko, E., Ceponiene, R., Balan, P., Fellman, V., & Naatanen, R. (2002). Maturation of the auditory change detection response in infants: a longitudinal ERP study. NeuroReport, 13, 1843–1848.Kutas, M., & Hillyard, S. (1980). Reading senseless sentences: Brain potentials reflect semantic incongruity. Science, 207, 203– 205.Leppaanen, P., Pihko, E., Eklund, K. M., & Lyytinen, H. (1999). Cortical responses of infants with and without a genetic risk for dyslexia: II group effects. NeuroReport, 10, 969–973.Mills, D., & Conboy, B. T. (2005). Do changes in brain organization reflect shifts in symbolic functioning? In L. Namy (Ed.), Symbol use and symbolic representation (pp. 123–153). Mahwah, NJ: Lawrence Erlbaum Associates.Mills, D. L., Prat, C., & Zangl, R. (2004). Language experience and the organization of brain activity to phonetically similar words: ERP evidence from 14- and 20-month-olds. Journal of Cognitive Neuroscience, 16, 1452–1464.Molfese, D. L. (2000). Predicting dyslexia at 8 years of age using neonatal brain responses. Brain and Language, 72, 238–245.Morr, M. L., Shafer, V. L., Kreuzer, J. A., & Kurtzberg, D. (2002). Maturation of Mismatch Negativity in typically- developing infants and pre-school children. Ear and Hearing, 23, 118– 136.Näätänen, R. (2001). The perception of speech sounds by the human brain as reflected by the mismatch negativity (MMN) and its magnetic equivalent (MMNm). Psychophysiology, 38, 1–21.Näätänen, R., Lehtokoski, A., Lennes, M., Cheour, M., Huotilainen, M., Livonen, A., et al. (1997). Language-specific phoneme representations revealed by electric and magnetic brain responses. Nature, 385, 432–434.Neville, H., Mills, D., & Lawson, D. (1992). Fractioning language: Different neural subsystems with different sensitive periods. Cerebral Cortex, 2, 244–258.Neville, H. J., Coffey, S. A., Holcomb, P. J., & Tallal, P. (1993). The neurobiology of sensory and language processing in language impaired children. Journal of Cognitive Neuroscience, 5, 235–253.Oberecker, R., Friedrich, M., & Friederici, A. D.[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... (2005). Neural correlates of syntactic processing in two-year-olds. Journal of Cognitive Neuroscience, 17, 1667–1678.Plante, E., Swisher, L., Vance, R., & Rapcsak, S. (1991). MRI findings in boys with specific language impairment. Brain and Language, 41, 52–66.Ponton, C. W., Eggermont, J. J., Kwong, B., & Don, M. (2000). Maturation of human central auditory system activity: Evidence from multi-channel evoked potentials. Clinical Neurophysi- ology, 111, 220–236. 54. 44 Valerie L. Shafer and Karen Garrido-NagPraamstra, P., Meyer, A., & Levelt, W. (1994). Neurophysiological manifestations of phonologi- cal processing latency variation of a negative ERP component timelocked to phonological mismatch. Journal of Cognitive Neuroscience, 6, 204– 219.Pulvermüller, F. (1996). Hebb’s concept of cell assemblies and the psychophysiology of word processing. Psychophysiology, 33, 317–333.Rivera- Gaxiola, M., Klarman, L., Garcia-Sierra, A., & Kuhl, P. (2005). Neural patterns to speech and vocabulary growth in American infants. NeuroReport, 16, 494– 498.Scott, S., & Wise, R. (2004). The functional neuroanatomy of prelexical processing in speech perception. Cognition, 92, 13–45.Shafer, V. L., Kessler, K. L., Morr, M. L., Schwartz, R. G., & Kurtzberg, D. (2005). Spatiotem- poral brain activity to “the” in discourse. Brain and Language, 93, 277–297.Shafer, V. L., Morr, M. L., Datta, H., Kurtzberg, D., & Schwartz, R. G. (2005). Neurophysio- logical indices of speech processing deficits in children with specific language impairment. Journal of Cognitive Neuroscience, 17, 1168– 1180.Shafer, V. L., Morr, M., Kreuzer, J., & Kurtzberg, D. (2000). Maturation of mismatch negativity in school-age children. Ear and Hearing, 21, 242– 251.Shafer, V. L., Schwartz, R. G., & Kessler, K. L. (2003). ERP indices of phonological and lexical processing in children and adults. In Proceedings of the 27th Annual Boston University Conference on Language Development (pp. 751–761). Somerville, MA: Cascadilla Press.Shafer, V. L., Schwartz, R. G., & Kurtzberg, D. (2004). Language specific memory traces of consonants in the brain. Cognitive Brain Research, 18, 242–254.Shafer, V. L., Schwartz, R. G., Morr, M. L., Kessler, K. L., & Kurtzberg, D. (2000). Deviant neurophysiological asymmetry in children with language impairment. NeuroReport, 11, 3715–3718.Shafer, V. L., Shucard, D. W., & Jaeger, J. J. (1999). Cerebral specialization and the role of prosody in language acquisition in three-month-old infants. Developmental Neuropsychology, 15, 73–110.Shucard, D., & Shucard, J. (1990). Auditory evoked potentials and hand preference in 6-month- old infants: Possible gender related differences in cerebral organization. Developmental Psychol- ogy, 26, 923–930.Squire, L., & Kandel, E. (1999). Memory: From mind and molecules. New York: Scientific Ameri- can Library/Scientific American Books.Strange, W. S., & Shafer, V. L. (in press). Speech perception in late second language learners: The re-education of selective perception. Cambridge: Cambridge University Press.Uwer, R., Albrecht, R., & von Suchodoletz, W. (2002). Automatic processing of tones and speech stimuli in children with specific language impairment. Developmental Medicine and Child Neurology, 44, 527–[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... 532.Vaughan, H., & Kurtzberg, D. (1992). Electrophysiologic indices of human brain maturation and cognitive development. In M. R. Gunnar, & C. A. Nelson (Eds.), Developmental behav- ioral neuroscience. Hillsdale, NJ: Lawrence Erlbaum Associates.Wagovich, S. A., & Newhoff, M. (2004). The single exposure: partial word knowledge growth through reading. American Journal of Speech Language Pathology, 13, 316–328.Wassenaar, M., & Hagoort, P. (2005). Word-category violations in patients with Broca’s aphasia: An ERP study. Brain and Language, 92, 117–137.Weber, C., Hahne, A., Friedrich, M., & Friederici, A. (2004). Discrimination of word stress in early infant perception: electrophysiological evidence. Cognitive Brain Research, 18, 149– 161. 55. The Neurodevelopmental Bases of Language 45Weber-Fox, C. M., & Neville, H. J. (1995). Maturational constraints on functional specializa- tions for language processing: ERP and behavioral evidence in bilingual speakers. Journal of Cognitive Neuroscience, 8, 231–256.Winkler, I., Kujala, T., Alku, P., & Naatanen, R. (2003). Language context and phonetic change detection. Cognitive Brain Research, 17, 833–844.Winkler, I., Kujala, T., Tiitinen, H., Sivonen, P., Alku, P., Lehtokoski, A., et al. (1999). Brain responses reveal the learning of foreign language phonemes. Psychophysiology, 36, 638–642. 56. 3Formal and Computational Constraints onLanguage DevelopmentHelen GoodluckThis chapter reviews approaches to change in children’s grammatical abilities from theperspective of work in the framework of generative grammar. We will evaluate the ideathat principles of grammar may mature over time, and look at the potential role of themechanisms for language production and comprehension in explaining children’s non-adult behaviors. Virtually all of the topics covered in this chapter are the subject ofongoing research and debate; my goal is simply to give a flavor of central issues, albeitwith a bias toward my own interests.Constraints on the Learner: Input Limitations andUniversal GrammarThis section considers what type of situation the learner confronts when acquiring a firstlanguage, and how the learner’s inbuilt knowledge of grammar may help him to copewith the limitations of the input he receives.Input constraintsThere are three major ways in which the linguistic input to the child is restricted. First,the child hears only a small subset of the infinite number of grammatical sentencesin the ambient language. Second, the set of sentences the child hears may not fully rep-resent the range of sentences that are grammatical in the adult language – some struc-tures are grammatical but rare. Third, the child is not corrected if he or she produces a 57. Formal and Computational Constraints 47sentence that is not grammatical in the adult language. This restriction is generally calledthe no negative evidence constraint. At least the third of these purported constraints on the input has been subject to agood deal of controversy from the 1970s onwards. However, I believe the state of currentevidence indicates that the observation that children are not instructed – either explicitlyor implicitly – is correct (see[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Valian, 1999, for a review).Constraints on hypothesis formation: Universal GrammarDespite the fact that the input to the learner is limited in the way just sketched, thesentences children are exposed to are in principle frequently amenable to multiple possi-ble analyses, including analyses that do not conform to the patterns found in adultgrammars. However, as we will see below, there is little evidence for children forminggrammars that deviate substantially from the patterns attested in adult languages. In1965 Noam Chomsky broke new ground, proposing a solution to the puzzle of howchildren achieve implicit knowledge of the grammar of their language in the face ofinput constraints. The first section of Aspects of the theory of syntax outlined the structureof a language acquisition device (LAD), proposing that the child tackled the task ofanalyzing the input with the use of an innate knowledge of the shape of human lan-guages: Universal Grammar (UG). Universal Grammar comprises formal universals andsubstantive universals. Formal universals limit what types of rules are allowed in humanlanguages. An elementary example is that no language uses linear order in the sentenceas a basis for signaling the difference between sentence types. The difference between adeclarative sentence and a yes–no question in English is signaled by inverting the subjectof the sentence and the first verb, not by, for example, inverting the order of the firsttwo words, as illustrated in (1) (an asterisk indicates ungrammaticality): (1) a. This sentence is declarative. b. Is this sentence declarative? (Inversion of subject and verb) c. *Sentence this is declarative? (Inversion of the order of first two words)Substantive universals are the “building blocks” of grammatical rules: the vocabularyout of which rules are constructed. Thus the categories Noun, Verb, Adjective, etc. (ora set of primitive features in terms of which these categories are defined) are substantiveuniversals. Implicit knowledge of the formal and substantive universals that compriseUG will give the child a head- start in analyzing the input, steering her away from amultitude of logically possible, but grammatically impossible, hypotheses about thestructure of her language. Chomsky’s 1981 book Lectures on government and binding put an important new spinon the concept of UG, by tackling language variation. Universal Grammar limits therange of possible human languages, but we don’t all speak the same language or type oflanguage. Not only are the vocabularies of different languages different, but so are thestructures. Languages such as English have the basic word order subject–verb– object(SVO); languages such as Japanese have the basic word order SOV; and languages such 58. 48 Helen Goodluckas Irish have the basic word order VSO. Languages such as English do not allow thesubject of a sentence to be omitted (2b); languages such as Italian permit the equivalentof such a sentence. (2) a. She left in a huff. b. *Left in a huff.Although it is possible to question a position inside an embedded clause (3a), languagessuch as English do not allow questions such as (3b), in which the question word refersto a position inside a relative clause. By contrast, languages such as Japanese or Akan(spoken[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... in Ghana) permit their equivalent. (3) a. What did Jane think that the man had eaten? b. *What did the man eat a veggieburger that contained? (cf. The man ate a veggieburger that contained tofu.)Chomsky’s basic proposal was that the principles of UG (formal universals) could be setto different values: the principles had parameters that needed to be fi xed by the languagelearner on the basis of the evidence of the language she hears.1Language Development: Continuity versus MaturationIn empirical work on language acquisition, the concepts of UG and principles andparameters were incorporated in the late 1970s onwards into hypotheses about the real-time course of language development. The most popular position was, and I believecontinues to be, that of continuity. Under continuity, the child’s developing grammar ishypothesized to be always a grammar that is a possible adult grammar, although it maynot be the correct grammar for the target language. For continuity to be maintained,given the assumption of no negative evidence, the input must contain clues that willpermit the child to deduce that she has not got the target grammar right, and hence thatshe must revise her grammar. In the terminology of principles and parameters, the inputmust be such that the child can reset a parameter to the correct value. Another approach to child grammars is maturation (Borer & Wexler, 1987). The ideais that the child is equipped innately with knowledge of all properties of UG, but thatsome principles and/or substantive universals may kick in only after a period of time, ina manner similar to the biologically determined schedule for physical growth. A view ofgrammatical development that admits maturation is consistent with continuity, since thechild may have enough in his arsenal of grammatical constructs to form a possible adultlanguage, even if he doesn’t have enough to construct the correct grammar of thelanguage he is learning (Borer & Wexler, 1992; Clahsen, 1990/91; Wexler, 1999). Agrammar that does not fit the target can be corrected by maturation of some principle(s).Such maturation may or may not be triggered by input, just as physical growth may bemore or less dependent on diet and other external factors. 59. Formal and Computational Constraints 49 The continuity and maturation views of development differ in the source of non-adultgrammars. Under maturation, the child forms a non-adult grammar because she doesn’thave available yet the apparatus to form the adult rules; under continuity, the child mayhave all of the constructs of UG at her disposal, but she does not deploy some types ofrule where the adult language does. The challenge for this view is to explain why thechild eschews certain grammatical mechanisms. The maturation and continuity viewsalso differ in that in principle maturation does allow for the child to form grammarsthat don’t fall within the range of adult grammars, although as just noted this is not anecessary corollary of maturation.Some Examples of Children’s Grammatical Knowledgeand DevelopmentThis section and the following one look at how the ideas just sketched fare in the faceof studies of particular aspects of grammatical development. By about 18 months chil-dren are producing their first multiword[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... utterances and there is substantial evidence thatthe parameter of basic word order is set very early, with consequences for other aspectsof grammar (see, e.g., Hickey, 1990; Hirsh-Pasek & Golinkoff, 1991; Otsu, 1994), andwith little evidence of experimentation or error. Other examples offer more evidence ofdevelopment.Example 1: Subjectless sentencesAs mentioned above, languages vary as to whether they permit subjectless sentences suchas (2b): Italian and Portuguese are among the languages that do, English is among thelanguages that do not. At early stages of language development, children learning English-type languages do produce subjectless sentences. Examples are given in (4) (taken fromthe speech of Adam (Brown, 1973), at age 27–29 months (cited in Pinker, 1995)): (4) play checkers screw part machine now put boots onAn early theory concerning such child utterances was that children learning Englishstart off with an Italian-type grammar and then have to reset the parameter for subject-less sentences to block this possibility in English (Hyams, 1986). However, cross-linguistic work has largely discredited this hypothesis. For example, in a study of2-year-old Portuguese-speaking children Valian and Eisenberg (1996) found that thechildren increased their use of subject pronouns over time. This argues that subjectomission by English-speaking children does not reflect a mis-set grammatical parameter,but is more likely a reflection of error due to pressure on the immature language produc-tion system (see Bloom, 1990, and Valian & Aubry, 2005, for further discussion). 60. 50 Helen GoodluckExample 2: Structural conditions on pronoun interpretationChomsky (1981) formulated a set of principles that govern the interpretation of definiteand reflexive pronoun interpretation. The definite pronoun her in (5a) may not refer toSue, but it may refer to Mary or to someone not mentioned in the sentence: (5) a. Mary realized that Sue had e- mailed her. b. Mary realized that Sue had e-mailed herself.By contrast, herself in (5b) must refer to Sue: it cannot refer to Mary. This basic contrastcan be accounted for by requiring the reflexive pronoun to refer to a noun (phrase) thatis structurally close to it. In the case of the English sentence (5b) “structurally close”means inside the lower, embedded sentence, not inside the main clause: (6) IP (Inflectional phrase = S(entence) in earlier analyses) NP VP Mary V CP (Complementizer phrase; host to words such as “that”) realized that IP NP had VP Sue V NP e-mailed her { herself }In the late 1970s and the 1980s, studies using a variety of experimental techniquesshowed that preschool children are sensitive to contrasts such as that in (5a) versus (5b),and that this sensitivity was based upon structural properties of the sentence, not uponsurface order of words (see Kaufman, 1994, for a review). Lidz (this volume) gives detailsof the structural relationship on which the possibility of co-reference for pronounsand reflexives is determined, and other examples of the contrasts to which children aresensitive. However, children’s performance is not error free. Many studies have shown that moremistakes are made with the interpretation of definite pronouns than with reflexives andthis has spawned a rich discussion[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... of factors which may affect children’s behavior. Defi-nite pronouns differ from reflexives in permitting extra-sentential (discourse) reference,and there are also discourse-related circumstances in which structural restrictions ondefinite pronoun interpretation may be overridden. For example, in the following dis- course the pronoun in speaker B’s utterance can be felicitously taken to refer to Mary,although this is not a permissible interpretation for the sentence taken in isolation: 61. Formal and Computational Constraints 51 (7) Speaker A: I don’t know anyone who likes Mary. Speaker B: Mary likes her.One approach to children’s errors with definite pronouns is to suggest that children mayhave trouble in executing such discourse-related conditions (see Chien & Wexler, 1990;Grodzinsky & Reinhart, 1993; Thornton & Wexler, 1999, among others). A second approach to children’s troubles with definite pronouns is to ask whetherchildren know which lexical items are reflexive and which are definite. Some languages,such as Danish, have a more complex system of reflexive and non-reflexive pronouns,with a three-way distinction: reflexives and definite pronouns such as those in English,plus a reflexive that is “long distance,” permitting the interpretation of sentences broadlyequivalent to (5b) in which herself is taken to refer to Mary. Other languages have a“simpler” system in which there is just one pronominal form, used for both reflexive andnon-reflexive meaning. An example is Maori. Given this cross-linguistic complexity, itis not unreasonable to suppose that the child might make errors that are rooted in mis-construing a definite pronoun in languages such as English as a reflexive lexical itemthat permits local reference (see Fodor, 1994, and Elbourne, 2005, for pertinentdiscussion). Moreover, children’s problems with pronouns are not restricted to definite pronouns.As just mentioned, Danish is a language that has a “long-distance” reflexive. Childrenlearning such languages will have to use the input (sentences they hear) to work out thatthis is a grammatical option. That input may not be readily available. Jakubowicz (1994)shows that children learning Danish may be as old as 9 or 10 years before they masterthe fact that in a sentence such as (8) the long-distance reflexive sig cannot refer to thesubject of the subordinate clause (Ida) but must refer to the subject of the main clause(Minnie). (This restriction is due to the nature of the verb in the embedded clause.) AsJakubowicz argues, the lengthy period of development for sig may be a reflex of the rarityof the sig construction in the input to the child. (8) Minnie beder Ida om at pege på sig. Minnie asks Ida point at self. “Minnie asks Ida to point at herself (=Minnie).”Example 3: MovementWe saw with examples (3a,b), repeated here as (9), that there are limits on the positionsin a sentence that can be questioned, questioning from within a relative clause beingblocked in English: (9) a. What did Jane think that the man had eaten? b. *What did the man eat a veggieburger that contained? Such constraints are associated with a mechanism that moves the question word fromits underlying position (as object of eat/contain in the examples) to the front of the[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... 62. 52 Helen Goodlucksentence. The gist of the linguistic account of the difference between (9a) and (9b) isthat there is an extra layer of structure in (9b), the noun phrase, and this prevents move-ment from within the relative clause: (10) a. Partial structure for (9a) (10) b. Partial structure for (9b) VP VP V CP V NP think that eat NP CP a veggieburger thatNor can we question a position within an embedded question (11) or a temporal clause(12): (11) *What did Dave wonder who wrote __? (cf. Dave wondered who wrote that memo.) (12) *What did Dave write that memo after he read __? (cf. Dave wrote that memo after he read Jane’s letter.) As mentioned above, not all languages use a movement mechanism and its attendantconstraints for question formation, presenting the learner with the challenge of workingout which language type he is exposed to (movement or non-movement). During the 1990s, several studies showed children to be sensitive to constraintson question formation in languages such as English. One of the clearest examples isde Villiers and Roeper (1995). They asked preschool children questions such as thosein (13): (13) a. How did the man who hurt his leg get home? b. How did the man rescue the cat who broke her leg?The questions were preceded by a short (two to three) sentence story that gave (for 13a)a manner of getting hurt and a manner of getting home and (for 13b) a manner of res-cuing and a manner of breaking a leg. In a question such as (13a), where the verb hurtis embedded within a relative clause modifying the man, how was never interpreted asreferring to the way in which the injury took place. By contrast, in a question such as(13b), where rescue is the main verb of the sentence, children freely construed the ques-tion word as referring to that verb (see also de Villiers, Roeper, & Vainikka, 1990, andGoodluck, Foley, & Sedivy, 1992, for evidence that children are sensitive to the imper- missibility of questions such as (11) and (12), respectively). While children have been shown to be sensitive to structural restrictions on questionformation associated with a movement mechanism, there is some debate about the degreeto which children’s earliest grammars employ movement in questions (see, e.g., discus- 63. Formal and Computational Constraints 53sion in Thornton & Crain, 1994). In the case of relative clause formation, it is quiteevident that children may go through a period of development before settling on theadult grammar. Just as in the case of question formation, there is cross-linguistic varia-tion in the mechanisms used to form relative clauses. Some languages, such as English,use a movement mechanism and the internal structure of relative clauses is subject tothe same structural constraints that govern question formation. Other languages, suchas Akan, do not use a movement mechanism and relative clauses are not subject tostructural constraints of the kind seen in English. In these languages, the noun phrasethe relative clause modifies is linked to a position inside the relative via a mechanism ofpronominal binding. Still other languages employ a dual mechanism for relative clauses,using both movement and binding. Such languages include[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Modern Irish and ModernHebrew. Even in English there are pronominal structures that permit linkage of anelement into a position from which movement is blocked. For example, in (14), (14) Bones, I know a dog that loves them. (cf. *What do I know a dog that loves?)bones refers to the object them inside the relative clause that loves them. The differencebetween languages such as English and languages such as Akan, Irish, and Hebrew isthat the latter type of language uses structures with pronominal binding much morefreely than English does – allowing the equivalent of the starred question in (14), or therelative clause in (15): (15) *I met the man that everyone knows who betrayed (him). Many studies have elicited relative clauses from preschool and young school-age child-ren. Particularly pertinent to this discussion are studies of French (Labelle, 1990),Serbo-Croatian (Goodluck & Stojanovic , 1996), and Modern Irish (Goodluck, ´Guilfoyle, & Harrington, 2006). Children’s relatives do not always conform to the dic-tates of the adult grammar. A generalization that has emerged is that children producerelatives with the properties of a binding mechanism, even when the adult language usesonly a movement mechanism. As we have seen in (14), one property of a binding mecha-nism is the use of pronouns. A pronoun occupies the site that the noun phrase the relativemodifies refers to, as illustrated in the adult Irish examples in (16); such pronouns areknown as resumptive pronouns. (16) a. an carr a dtiomáineann Nell ar scoil gach lá é the car COMP drives Nell to school every day it “the car Nell drives to school every day” b. an fear a n-insíonn tú an scéal do the man COMP told you a story to-him “the man you told a story to”Labelle (1990) found her French- speaking subjects innovated resumptive pronouns,although these are ungrammatical in the adult language. Examples from Labelle’s dataare given in (17): 64. 54 Helen Goodluck (17) a. la petite fille qu’a est assis sur la boîte the little girl COMP she is sat on the box (NB: qu’a ([ka]) = que elle, “that she”) “the little girl that is sitting on the box” b. sur la balle qu’i(l) l’attrape on the ball COMP he it catches “on the ball that he catches” c. sur la boîte que le camion rentre dedans on the box that the truck goes inside-it “on the box that the truck is going inside” It might be supposed that the examples from child French in (17) are “slips of thetongue,” rather than a genuine stage in grammatical development. In speech, resumptivepronouns are sometimes inserted in languages that do not permit resumptives, in orderto “repair” an ungrammatical structure. Such repairs are found particularly when thedistance between the resumptive and the element it refers to increases; compare (18a)with (18b,c): (18) a. *What did Jane eat a veggieburger that contained it? b. ?*What did Sue claim that Jane had eaten a veggieburger that contained it? c. ??What did Ida believe that Sue had claimed that Jane had eaten a veggieburger that contained it? The relatives in Labelle’s study and other studies cited above were elicited fromchildren in picture description tasks; perhaps the difficulty of this task promptsspeech errors, including repair resumptive pronouns. The body of evidence fromthe various studies argues[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... against this view as a complete account of the data. Forexample, another index of use of a binding as opposed to movement strategy for relativeclause formation is the use of an invariant complementizer such as English that tointroduce the relative clause. The children in Labelle’s study used the invariant com-plementizer que to introduce their relatives, although the adult grammar requires arelative pronoun; children acquiring Serbo-Croatian produced relative clauses with aninvariant complementizer (as opposed to a relative pronoun) more frequently thanadults did (Goodluck & Stojanovic , 1996). And children learning Irish innovated a ´structure that is plausibly the result of use of binding to form a relative clause inan environment where binding relatives are blocked in the adult grammar (Goodlucket al., 2006). As in the case of subjectless sentences, we can appeal to pressure on the productionmechanism to account for children’s innovation of binding relatives. On the assumptionthat the movement operation adds an extra computational burden in producing a rela-tive, we can say that children innovate binding relatives for ease of production. Notethat in this case, production pressure leads to the temporary positing of a non-adultgrammar, whereas in the case of subjectless sentences, production pressure leads to non-adult utterances that are in fact also ungrammatical for the child. 65. Formal and Computational Constraints 55SummarySo far we have seen that from an early age children are sensitive to grammatical restric-tions present in the ambient language, including whether their language permits subject-less sentences, and restrictions on reflexive and definite pronoun interpretation. Whenchildren do err, we have appealed to a variety of factors: pressure on the language pro-duction system (as in the case of English- speaking children’s subjectless sentences andthe innovation of a binding mechanism for forming relative clauses); the need to accessdiscourse/real- world knowledge; the need to correctly identify grammatical categories(which elements are definite pronouns and which are reflexive); and the need to accessinput that is rare (Danish long-distance reflexive sentences). In each case it is possibleto maintain continuity, in the sense that the child does not posit a grammar that deviatesfrom what is permitted in adult languages. However, not all cases of non-adult behaviorcan easily be written off in a similar manner. In the next section, we will look at a casewhere children’s grammar appears to be in violation of continuity.A Problem for Continuity: Root InfinitivesOne property of adult languages that is a candidate for an inviolable, non-parameterizedconstraint of Universal Grammar is the fact that (a small number of special cases aside)main clauses are always tensed – i.e. marked for past or present (or future in many lan-guages). Thus it has generally been assumed that adult languages with sentences such asthat in (19) are not attested (though this may be too strong; Borer & Rohrbacher, 2002): (19) *He (to) sleep.But there is now abundant evidence that children produce such sentences. English hasa relatively impoverished morphological system for marking tense, with the infinitive(non-tensed) form of the verb being identical in many cases to the tensed from. But inlanguages such as[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Dutch or French, in which the infinitive and tensed forms are clearlydistinct, children’s use of infinitives in main clauses is plain to see. Hoekstra and Hyams(1998) studied root (main clause) infinitives produced by children in the age range ofapproximately 18 to 36 months; (20) gives examples (taken from Hoekstra & Hyams,1998, p. 83)2 : (20) a. Papa schoenen wassen Daddy shoes wash-inf. (Dutch; Weverink, 1989) b. Michel dormi Michel sleep-inf. (French; Pierce, 1992) c. Thorstn das haben Thorstn that have-inf. (German; Poeppel & Wexler, 1993) d. Jag också hoppa där å där I also hop-inf. there and there (Swedish; Santelmann, 1995) 66. 56 Helen GoodluckIf adult grammars impose the restriction that main clauses must be tensed, then suchexamples are a problem for continuity, which claims that children’s grammars are alwayspossible adult grammars, even if these child grammars do not always correspond to thegrammar of the language at hand. There is now a large literature on root infinitives, some of it advocating maturationand some continuity. Cross-linguistic comparisons have revealed consistent patterns inthe distribution of children’s root infinitives, leading the way for the development of anaccount of their distribution in which the deviation from adult grammars is not as largeas it might at first appear. The following is based on Hoekstra and Hyams (1998). Two major generalizations concerning child root infinitives in languages such asDutch are the following:(a) Root infinitives are associated with eventive verbs (such as wash or hop) rather than stative verbs (such as be or love) (Ferdinand, 1996; Jordens, 1991). Eventive verbs basically denote actions or events, and stative verbs denote states/conditions.(b) Root infinitives are associated with modal interpretations – specifically, an inter- pretation of necessity or desire, as illustrated in the following child Dutch examples (based on Hoekstra & Hyams, p. 92). (21) a. Eerst kaartje kopen! First ticket buy-inf. Intended meaning: “We must first buy a ticket” b. Niekje buiten spelen Niekje outside play-inf. Intended meaning: “Niek (=speaker) wants to play outside” Hoekstra and Hyams propose an analysis along the following lines. The infinitive inadult languages has a particular aspectual property: it “denotes an event not yet realized”(Hoekstra & Hyams, p. 102). Hoekstra and Hyams point out that this makes sense ofthe fact that in French, for example, the future tense is represented by an affi x that isadded to the infinitival form (j’arriver-ai “I shall arrive,” where arriver is the infinitiveand ai the future affi x). Returning to the fact mentioned above, that adult languages dodisplay root infinitives in a small number of special cases, we can see that these specialcases also involve events not yet realized. The Dutch examples in (22) (from Wijnen,1996, quoted in Hoekstra & Hyams, p. 103) illustrate this property: (22) a. Hier geen fietsen plaatsen! Here no bicycles place-inf. “Don’t put bicycles here!” b. Jan met mijn zus trouwen?! Dat nooit. Jan my sister marry- inf. That never.So now we are seeing a similarity between child root infinitives and adult grammars.But the fact that children produce so many such utterances in situations where the adultgrammar requires other means of expressing the intended meaning (such as the use of[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... 67. Formal and Computational Constraints 57a modal verb) still requires an explanation. Hoekstra and Hyams propose that wherechildren and adults differ is in the degree of freedom they are allowed in interpreting asentence with reference to the discourse. Roughly, they propose that adults are largelyconstrained to provide an interpretation of the sentence within its syntactic structure:the syntactic elements that represent tense must be fully interpreted inside the sentence.By contrast, children may use the discourse to “fill in” information about tense that isnot computed internally to the syntax.Processing ConstraintsThus far we have been concerned with constraints on children’s hypothesis formationthat derive from Universal Grammar, although we have seen that the outer course ofdevelopment may be affected by the capacity of the sentence production mechanism.This section deals with another potential capacity limitation, that of sentence processing.There has been little intersection between research of language processing and languageacquisition, despite the importance of the former for the latter – the child has to processthe input to form a grammar (see Fodor, 1998, for one exception to this lack ofintersection).The sentence processorThe model of sentence processing that we assume for adults is one in which grammaticalstructure is built immediately as the sentence is input, and that much, if not all, of thedetail that is justified in the theoretical literature is present in the syntactic structuresassembled by the processor. In addition, we will assume the following: (a) small frag-ments of syntactic structure (phrasal chunks) are built immediately; (b) these chunksare subsequently assembled into larger structures, which may be partially determined ina top- down manner; (c) sentences (clauses) are important processing units, and sentenceboundaries are the point at which material is recoded into a non- immediate memoryrepresentation, which may lose much of the fine-grained syntactic structure of first-stageprocessing; (d) the integration of sentential structure with a larger discourse-orientedrepresentation takes place at a relatively late stage in the chain of processing operations;(e) the processor is a capacity-limited device that can overload. Figure 3.1 illustrates thisview of sentence processing. Some of these assumptions are more controversial than others. For example, (c) issupported by extensive experiments from the 1960s onwards, whereas (d) is the subjectof considerable debate. See Fodor (1995), Tanenhaus and Trueswell (1995), and Treiman,Clifton, Meyer, and Wurm (2003) for useful reviews of the literature, and Frazier (1999)and Gibson (1998) for more technical discussions. We will assume that the child’s processing device has the same architecture as theadult’s. Aside from a different stock of first language rules to draw on, the differencebetween children and adults with respect to processing will boil down to a difference in 68. 58 Helen Goodluck WORLD KNOWLEDGE UNIVERSAL GRAMMAR/ AND COMPETENCE GRAMMAR DISCOURSE SITUATION M Initial lexical Sentence-level Construction of E Input and syntactic integration of discourse S analysis syntactic units representation S A G EFigure 3.1 The human[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... sentence processing mechanism.processing capacity: the child’s working memory and storage capacity will be less thanthe adult’s. What consequences will this lesser processing capacity have for language develop- ment? Newport (1988) has suggested that less may be more – if the child takes in onlysmall chunks of language structure, this may allow her to analyze the details of the input,establishing morphological boundaries that may be crucial to cracking the language’ssyntactic system. On the other hand, lesser processing capacity may lead to error. Con-sider again the case of definite pronouns. Although preschool children show knowledgeof structural restrictions on definite pronoun reference, errors are committed. Oneexplanation we sketched above for this was that children may falter in their knowledgeand/or execution of pragmatic, discourse-related principles that enter into definitepronoun interpretation, above and beyond structural restrictions. The model in Figure3.1 is complementary to this account, and may be argued to explain children’s difficultywith discourse-related operations: if children have lesser processing capacity and hencelesser ability to perform operations towards the end of the chain of operations in Figure3.1, they will be more likely to make errors with definite pronouns than with reflexives(Goodluck, 1990; see also Avrutin, 2000), since the former but not the latter may requireaccess to multiclause structures and discourse information. In favor of a processingaccount of some if not all errors with pronoun interpretation is the fact that there isevidence that children make errors with pronoun interpretation that can be linked todifficulty in accessing discourse, and which do not fall under other approaches to errorswith pronouns (see, e.g., Goodluck & Solan, 2000; Goodluck, Terzi, & Chocano Díaz,2001). Another angle on the way in which the sentence processing mechanism may guidedevelopment concerns the issue of no negative evidence. It can be argued that the proces-sor leads the child to conservative hypotheses, preventing her from positing incorrectgrammars for which there will be no corrective input. Consider again the examples in(13). English is a language that does not permit a question word to refer inside a relativeclause – hence how in (13a) cannot be taken to modify the verb hurt. De Villiers andRoeper (1995) showed that children categorically obeyed this restriction. However, thereis another explanation for their findings. One of the properties of the sentence processingmechanism listed above is the importance of clausal units. Suppose the processor oper-ates under the following constraint: 69. Formal and Computational Constraints 59 (23) As the sentence is input, link a question word to a position in the incoming string only when the link results in a complete proposition.3The fact that hurt is contained within a relative clause in (13a), whereas the verb rescuein (13b) is not, means that (23) will be violated if how is linked to the first verb in thesentence in the case of (13a), but not in the case of (13b): (24) a. How did the man who hurt his leg . . . . . . . (incomplete proposition) b. How did the man rescue the cat . . . . . . . . . (complete proposition)It may seem that the processing constraint[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... (23) simply reduces to knowledge of thegrammatical constraint on questioning from within a relative clause: How can you knowthat the fragment in (24a) is not a complete proposition unless you have registered thefact that hurt is contained within a relative clause? However, these two accounts can bedistinguished. If the constraint in (23) is a general constraint on processing, we predictthat in a language which, unlike English, permits a question word to refer inside a rela-tive clause, in sentence processing (23) will still block the location of the question wordinside the relative clause in sentences equivalent to (13a). The relevant evidence has stillto be gathered, but there are some preliminary data that suggest this may be the case(Saah & Goodluck, 1995).Processing and learnabilityDuring the 1980s, there was considerable discussion of the problem of learnability. Itwas reasoned that if the child receives no negative evidence, the language acquisitiondevice must be structured in such a way as to prevent the child from hypothesizinggrammars that were incorrect, and too permissive in terms of the range of grammaticalsentence types sanctioned in the target language. Even if such a child grammar waspermitted under the constraints of Universal Grammar, the child would have no inputthat allowed him to correct his erroneous hypothesis. One solution to this problem wasthe subset principle (Berwick, 1985). The subset principle proposes that the child positsthe grammar that is (a) compatible with the input data (the speech he hears) and (b)generates the smallest set of output sentences. Thus, for example, suppose the childlearner hears a sentence such as (the equivalent of) (13b), repeated here: (25) How did the man rescue the kitten who broke her leg?Universal Grammar will tell him that there are two possibilities: either the question wordhow can refer only to the manner of rescuing (i.e., he is learning a language such asEnglish, which blocks a question word from referring to a position inside a relativeclause), or how can modify either verb (rescue, break). The subset principle will tell himto hypothesize that he is learning an English-type language, since that language typepermits the smaller range of possible interpretations. If the child is in fact learning alanguage that permits both interpretations, he will eventually receive input that tells him 70. 60 Helen Goodluckthat his grammar needs to be revised to allow a question word to refer inside a relativeclause – such input would be, for example, the equivalent of a question such as (3/8b)(What did the man eat a veggieburger that contained?). One criticism of the subset principle is that it is a principle for which there is littlein the way of independent evidence that it is operative as a cognitive and/or gram-matical principle. The proposal in the last section – that the processing mechanismmay steer the hearer away from some logically possible (and in some languagesgrammatically permissible) interpretations – has the potential to explain cases wherethe child opts for a less liberal grammar than his language in fact allows, and to thusdo away with the subset principle as a principle governing the language acquisitiondevice.More on Continuity versus MaturationI have just suggested that an understanding of sentence processing limitations may helpus do away[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... with an otherwise stipulative principle guiding the child’s hypothesis forma- tion. In this section we will look again at root infinitives and also at passive sentences,and consider whether performance limitations – limitations on the child’s ability toexecute her knowledge in real-time production and comprehension – can replace theidea that grammatical principles may mature.Root infinitives revisitedThere is superficially a tension between the claim that children may have difficultyaccessing discourse representations (contributing to their difficulties with definitepronoun interpretation) and the analysis of early root infinitives as a result of children’sgreater reliance on discourse to anchor the sentence in time relations. This is not neces-sarily a conflict. Notice that the difficulty children have with pronouns is to do withcomprehension, whereas in the case of root infinitives we are dealing with production.We have already suggested that children’s non-adult use of subjectless sentences andchildren’s predilection for non-adult relative clauses formed by binding may be due topressure on the production system. A similar kind of analysis can be proposed in thecase of root infinitives. It is a fact that children’s root infinitives exist side by side withcorrect tensed forms. Thus children at the stage of producing root infinitives cannot besaid to lack knowledge of the realization of tense in their language. Since the late 1980s inflectional phrase (IP) in Chomskyan generative grammar hasbeen replaced by a series of “functional” categories, amongst which are categories thathost the representation of tense and aspect, and number (singular, plural) of the subject.Such functional categories provide an infrastructure for the semantics of the sentence.4Thus a sentence such as “John is eating sushi” will have a structure roughly along thelines shown in (26): 71. Formal and Computational Constraints 61 (26) FC1 (FC1, FC2, and FCn are functional categories containing representation of tense, aspect, number) FC2 FCn VP V NP eat sushiThis infrastructure of functional categories is critical to the representation of tense in adultlanguages. In Hoekstra and Hyams’ (1998) analysis, there is an abstract tense operator (TO)in a complementizer phrase that binds the tense position (realized in TP = tense phrase): (27) TOi F1 . . . . . Fn Tensei VPThis binding operation is critical to the meaning of the sentence. Now consider the situ-ation facing a child speaker of a language such as Dutch who wishes to express theproposition “We must first buy a ticket” (example 21a above). The child has an implicitchoice – either she can compute the functional structure in the same way as adults do,or she can draw on her knowledge of the fact that infinitive forms denote events not yetrealized, a fact that Hoekstra and Hyams argue is related to modal meaning, and producea root infinitive. If the child takes the latter option, she is saved the “bother” of comput-ing the tense operator structure for the sentence, but must rely on the discourse situationfor the meaning of the sentence to be complete. Seen in this way, root infinitives mayhave their origin in a limitation in the capacity of the production mechanism, that is,at an early stage children may lack the computational power to consistently build higherlevels of functional structure (cf. Rizzi, 1994)[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... and tense operator representations. Thusin this case the child’s lesser performance capacity in production may drive her into usingdiscourse to complete the representation of the sentence, just as in the case of pronouninterpretation the structure of the processing device may deny her adequate access todiscourse.5 (See Borer & Rohrbacher, 2002, for a more recent analysis of root infinitiveswhich draws on literature demonstrating reliance on discourse to specify tense in adultlanguages. Borer and Rohrbacher propose that children’s root infinitives do involve theprojection of (phonetically null) functional categories; however, their account is notincompatible with the account sketched above, given that the levels of functional cate-gory needed for discourse interpretation of tense are fewer than those needed for thesentence-internal specification of tense.)Movement in passive sentencesDebate about whether children innovate a non- movement mechanism where the adultlanguage uses a movement mechanism is not confined to the acquisition of relative 72. 62 Helen Goodluckclauses. In fact, the use of movement with respect to the development of passive sentenceswas a major impetus to the idea that some properties of grammar may mature in thecourse of childhood. In languages such as English, passive sentences are formed bymoving the direct object of the sentence to subject position, as illustrated in (28): (28) a. [e] was eaten the cake (by Billy) (structure before movement) b. The cakei was eaten [ei] (by Billy) (structure after movement) ([e] indicates an empty phrase, to which the object is moved.)A number of studies have shown that preschool children have more trouble understand-ing passive sentences when they contain a non-action verb, such as love, than when theycontain an action verb, such as eat. Borer and Wexler (1987) proposed that this asym- metry could be explained if children lacked the ability to perform movement to subjectposition (or whatever grammatical properties motivate that operation), but rather formedpassives with the subject of the passive sentence always in subject position. They arguedthat the passive verb form (such as “eaten”) in the passives produced by children was infact an adjective, and that the difficulty children have with non-action passives is dueto the fact that non- action verbs cannot readily be converted into adjectives. Borer andWexler proposed that the capacity to perform movement to subject position matured inthe course of development. A number of objections to this proposal were raised in thelate 1980s and early 1990s (see Demuth, 1989; Pinker, Frost, & Lebeaux, 1987), and asignificant experimental advance was made by Fox and Grodzinsky (1998). They foundthat performance on non-actional passive sentences improved when the passive sentencedid not contain a by phrase. Thus sentences such as (29a) were found to be easier thansentences such as (29b), but the same was not true for (30a) versus (30b): (29) a. The cake was eaten by Billy. b. The thief was seen by Billy. (30) a. The cake was eaten. b. The thief was seen.Fox and Grodzinsky argue that this result can be accounted for if children’s difficultywith non-actional passives lies not with the operation of moving a direct object intosubject position, but rather with the[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... nature of the object of by. In (29a), Billy has thesemantic role of agent, whereas in (29b), Billy has the role of experiencer (the personwho undergoes the experience of seeing). They proposed that the agent role is the normal,“default” role conveyed by by, and that children’s difficulties with non- agentive passiveswith a by phrase come from the need to transmit the non- canonical semantic role ofexperiencer via the by phrase. When this need is eliminated, as in (30b), performanceon the non-agentive passives is improved. If this account is correct, then there is no needto propose that the ability to move to subject position matures.6Can computational limitations replace maturation?Maturational proposals have been made with respect to a number of areas of grammarin addition to the passive example just given, including movement in relative clauses(Guasti & Shlonsky, 1995) and the interpretation of adverbial clauses with missing sub- 73. Formal and Computational Constraints 63jects (Wexler, 1991; evaluated in Goodluck, 2001). Maturation handles non-adult behav-ior in a straightforward way: the child forms a non-adult grammar because she doesn’thave the ability to do otherwise. One potential problem for maturation is that the ageat which maturation is claimed to take place does not always jibe with evidence thatchildren do command the relevant operation/property of grammar. For example, innova-tion of non-adult relative clauses can persist until the child is age 6 or older, yet there isevidence that younger children do have a movement operation for forming relativeclauses (Goodluck et al., 2006; McKee & McDaniel, 2001). This problem is not serious.Wexler (1999) observes maturation can take place on a slightly different time schedulefor different children; moreover, even after a grammatical property has matured the childmay not be prompted to use it in her grammar immediately. The challenge for advocates of continuity without maturation is to find a motivationfor non-adult behaviors that does not deny children full access to Universal Grammar. Ibelieve that performance limitations of the kind we have sketched above may provide someanswers. We suggested that difficulty with definite pronoun interpretation may be rootedin relatively late access to discourse information by the sentence processing mechanism,and that a binding mechanism for relative clauses may be preferred because it is compu-tationally simpler than movement. If it is the case that children use an adjectival structurefor passives at early stages (modulo the reservations sketched above), then this also mayresult from the computational ease of a non-movement mechanism. Two objections, atleast, can be made to the general idea that children’s choice of a non-adult grammar maybe rooted in performance limitations. First, it may be said that the idea provides no expla- nation of why a particular mechanism is used by the child in one circumstance but notanother. For example, why do children avoid movement in relative clauses, but not indirect questions? Although the answer to such questions may not always be obvious, I feelthis is just a call for more thought on the way in which the human sentence processing/production mechanisms shape grammar and influence the cross-linguistic frequency ofstructures. In[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... the case of the asymmetry between questions and relative clauses, it appearsthat cross-linguistically a non-movement mechanism may have a special status for relativeclauses, something that Goodluck et al. (2006) suggest may be rooted in pressure on theproduction device. The second objection is that we know too little about the structureand capacity of the performance mechanisms to evaluate the claim that children’s abilityis quantitatively less than adults’. It is certainly the case, as noted above, that the organiza-tion of the processing mechanism is highly controversial, and the idea that access to dis-course takes place at a relatively late stage may not go down well in some research circles.However, progress is being made in evaluating children’s processing capacity and its rela-tion to the analysis of specific structures (Felser, Marinis, & Clahsen, 2003).Summary and ConclusionWe can summarize the facts of development sketched in this chapter as follows:• From early on, the child shows sensitivity to basic properties of the language- particular grammar he is learning (basic word order, subjectless sentences, structural constraints on pronominal reference). 74. 64 Helen Goodluck• Some structures elude the child for a lesser or greater period: root infinitives disap- pear fairly rapidly, but complete mastery of, inter alia, adult-like relative clause structures and rare constructions such as the Danish sig reflexive may take into the school years.It seems that the non-adult behaviors observed can be assimilated into a model of acquisi-tion that assumes continuity (with full access to Universal Grammar), although this isa matter of ongoing debate. A promising avenue for an account of development is theintegration of models of human sentence processing and production into models of lan-guage acquisition. Such an integration has the potential to supplant a previously pro-posed principle of language acquisition – the subset principle – as well as the constructof maturation, explaining changes in grammar over time in terms of an interactionbetween input, innate linguistic knowledge, and increased performance capacity.NotesMarilyn Shatz provided very helpful comments on both the form and content of earlier versionsof this chapter.1 Not touched on in this chapter is the influence on the study of language development of the Minimalist Program of Chomsky (1993, 1995) and subsequent work. This is because, to my understanding, this work has not to date had the impact on the study of language acquisition of the earlier frameworks (Aspects of the theory of syntax and Lectures on government and binding).2 The sources are as cited in Hoekstra and Hyams (1998).3 This general hypothesis may need to be modulated by capacity limitations (cf. J. Sedivy, unpublished master’s thesis, University of Ottawa, 1991; Aoshima, Phillips, & Weinberg, 2004); however, it has a body of evidence in its favor (Goodluck, Finney, & Sedivy, 1992).4 A large part of the continuity–maturation debate over the past 15 years has concerned whether functional categories mature. Guilfoyle and Noonan (1992) and Radford (1990) were early advocates of maturation; the more recent literature drawn on here is compatible with continuity.5 The account just sketched does not in and of itself explain the fact that in Hoekstra and Hyams’ (1998) survey root[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... infinitives are confined to languages in which tense is realized by number (singular, plural) marking for subject on the verb.6 Fox and Grodzinsky’s (1998) interpretation of their data and the issue of the development of the passive is a matter of ongoing debate (Babyonshev, Ganger, Pesetsky, & Wexler, 2001; Wexler, 2004).ReferencesAoshima, S., Phillips, C., & Weinberg, A. (2004). Processing filler-gap dependencies in a head- final language. Journal of Memory and Language, 51, 23–54.Avrutin, S. (2000). Comprehension of discourse-linked and non-discourse-linked questions by children and Broca’s aphasics. In Y. Grodzinsky, L. Shapiro, & D. Swinney (Eds.), Language and brain: Representation and processing. San Diego, CA: Academic Press. 75. Formal and Computational Constraints 65Babyonshev, M., Ganger, J., Pesetsky, D., & Wexler, K. (2001). The maturation of grammatical principles. Linguistic Inquiry, 32, 1–44.Berwick, R. (1985). The acquisition of syntactic knowledge. Cambridge, MA: MIT Press.Bloom, P. (1990). Subjectless sentences in child language. Linguistic Inquiry, 21, 491–504.Borer, H., & Wexler, K. (1987). The maturation of syntax. In T. Roeper & E. Williams (Eds.), Parameter setting. Dordrecht, The Netherlands: Reidel.Borer, H., & Wexler, K. (1992). Bi-unique relations and the maturation of grammatical princi- ples. Natural Language and Linguistic Theory, 10, 147–189.Borer, H., & Rohrbacher, B. (2002). Minding the absent: Arguments for the full competence model. Language Acquisition, 10, 123–175.Brown, R. (1973). A first language. Cambridge, MA: Harvard University Press.Chien, Y.-C., & Wexler, K. (1990). Children’s knowledge of locality conditions in binding as evidence for the modularity of syntax and pragmatics. Language Acquisition, 1, 225– 295.Chomsky, N. (1965). Aspects of the theory of syntax. Cambridge, MA: MIT Press.Chomsky, N. (1981). Lectures on government and binding. Dordrecht, The Netherlands: Foris.Chomsky, N. (1993). A minimalist program for linguistic theory. In K. Hale & S. J. Keyser (Eds.), The view from Building 20. Cambridge, MA: MIT Press.Chomsky, N. (1995). The minimalist program. Cambridge, MA: MIT Press.Clahsen, H. (1990/91). Constraints on parameter setting: A grammatical analysis of some acqui- sition stages in German child language. Language Acquisition, 1, 361–391.De Villiers, J., & Roeper, T. (1995). Relative clauses are barriers to wh-movement in young children. Journal of Child Language, 22, 389–404.De Villiers, J., Roeper, T., & Vainikka, A. (1990). The acquisition of long distance movement rules. In L. Frazier & J. de Villiers (Eds.), Language acquisition and language processing. Dor- drecht, The Netherlands: Kluwer Academic Publishers.Demuth, K. (1989). Maturation and the acquisition of the Sesotho passive. Language, 65, 56–80.Elbourne, P. (2005). On the acquisition of principle B. Linguistic Inquiry, 36, 333–365.Felser, C., Marinis, T., & Clahsen, H. (2003). Children’s processing of ambiguous sentences: A study of relative clause attachment. Language Acquisition, 11, 127–163.Ferdinand, A. (1996). The acquisition of the subject in French. PhD dissertation, HIL/Leiden University.Fodor, J. D. (1994). How to obey the subset principle: Binding and locality. In B. Lust, G.[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Hermon, & J. Kornfilt (Eds.), Syntactic theory and first language acquisition: Cross-linguistic perspectives. Vol. 2: Binding, dependencies and learnability. Hillsdale, NJ: Lawrence Erlbaum Associates.Fodor, J. D. (1995). Comprehending sentence structure. In L. Gleitman & M. Liberman (Eds.), An invitation to cognitive science (2nd ed.). Vol. 1: Language. Cambridge, MA: MIT Press.Fodor, J. D. (1998). Parsing to learn. Journal of Psycholinguistic Research, 27, 339–374.Fox, D., & Grodzinsky, Y. (1998). Children’s passive: A view from the by-phrase. Linguistic Inquiry, 29, 311–332.Frazier, L. (1999). On sentence comprehension. Dordrecht, The Netherlands: Kluwer Academic Publishers.Gibson, E. (1998). Linguistic complexity: Locality of syntactic dependencies. Cognition, 68, 1–76.Goodluck, H. (1990). Knowledge integration in processing and acquisition. In L. Frazier & J. de Villiers (Eds.), Language acquisition and language processing. Dordrecht, The Netherlands: Kluwer Academic Publishers. 76. 66 Helen GoodluckGoodluck, H. (2001). The nominal analysis of children’s interpretation of adjunct PRO clauses. Language, 77, 271– 287.Goodluck, H., Finney, M., & Sedivy, J. (1992). Completeness and filler- gap dependency parsing. In P. Coopmans, B. Schouten, & W. Zonneveld (Eds.), The OTS Yearbook (pp. 19–32). Utrecht, The Netherlands: Institute for Research in Language and Speech, University of Utrecht.Goodluck, H., Foley, M., & Sedivy, J. (1992). Adjunct islands and acquisition. In H. Goodluck & M. Rochemont (Eds.), Island constraints: Theory, acquisition and processing. Dordrecht, The Netherlands: Kluwer Academic Publishers.Goodluck, H., Guilfoyle, E., & Harrington, S. (2006). Merge and binding in child relative clauses: Evidence from Irish. Journal of Linguistics, 42, 629–661.Goodluck, H., & Solan, L. (2000). Un effet du principe C chez 1’enfant francophone. Canadian Journal of Linguistics/Revue canadienne de linguistique, 45, 49– 62.Goodluck, H., & Stojanovic, D. (1996). The structure and acquisition of relative clauses in ´ Serbo-Croatian. Language Acquisition, 5, 285– 315.Goodluck, H., Terzi, A., & Chocano Díaz, G. (2001). The acquisition of control cross- linguistically: lexical and structural factors in learning to license PRO. Journal of Child Language, 28, 158–172.Grodzinsky, Y., & Reinhart, T. (1993). The innateness of binding and coreference. Linguistic Inquiry, 24, 69– 101.Guasti, T., & Shlonsky, U. (1995). The acquisition of French relative clauses reconsidered. Lan- guage Acquisition, 4, 257–276.Guilfoyle, E., & Noonan, M. (1992). Functional categories and language acquisition. Canadian Journal of Linguistics/Revue canadienne de linguistique, 37, 241–272.Hickey, T. (1990). The acquisition of Irish: A study of word order development. Journal of Child Language, 17, 17–41.Hirsh-Pasek, K., & Golinkoff, R. (1991). Language comprehension: A new look at some old problems. In N. Krasnegor, D. Rumbaugh, R. Schiefelbusch, & M. Studdert-Kennedy (Eds.), Biological and behavioral determinants of language development. Hillsdale, NJ: Lawrence Erlbaum Associates.Hoekstra, T., & Hyams, N. (1998). Aspects of root infinitives. Lingua, 106, 81–112.Hyams, N. (1986). Language acquisition and the theory of parameters. Dordrecht, The Netherlands:[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Reidel.Jakubowicz, C. (1994). Reflexives in French and Danish: Morphology, syntax and acquisition. In B. Lust, G. Hermon, & J. Kornfilt (Eds.), Syntactic theory and first language acquisition: Cross-linguistic perspectives. Vol. 2: Binding, dependencies and learnability. Hillsdale, NJ: Law- rence Erlbaum Associates.Jordens, P. (1991). The acquisition of verb placement in Dutch and German. Linguistics, 28, 1407–1448.Labelle, M. (1990). Predication, wh- movement and the development of relative clauses. Language Acquisition, 1, 95–119.Kaufman, D. (1994). Grammatical or pragmatic: Will the real Principle B please stand up? In B. Lust, G. Hermon, & J. Kornfilt (Eds.), Syntactic theory and first language acquisition: Cross- linguistic perspectives. Vol. 2: Binding, dependencies and learnability. Hillsdale, NJ: Lawrence Erlbaum Associates.McKee, C., & McDaniel, D. (2001). Resumptive pronouns in English relative clauses. Language Acquisition, 9, 113–156.Newport, E. (1988). Constraints on learning and their role in language acquisition: Evidence from American Sign Language and language learning. In W. A. Collins (Ed.), Aspects of the 77. Formal and Computational Constraints 67 development of competence. Minnesota Symposia on Child Psychology: Vol. 14. Hillsdale, NJ: Lawrence Erlbaum Associates.Otsu, Y. (1994). Case marking particles and phrase structure in early Japanese acquisition. In B. Lust, M. Suñer, & J. Whitman (Eds.), Syntactic theory and first language acquisition. Vol. 1: Heads, projections and learnability. Hillsdale, NJ: Lawrence Erlbaum Associates.Pierce, A. (1992). Language acquisition and syntactic theory: A comparative analysis of French and English child grammars. Dordrecht, The Netherlands: Kluwer Academic Publishers.Pinker, S. (1995). Language acquisition. In L. Gleitman & M. Liberman (Eds.), An invitation to cognitive science (2nd ed.). Vol. 1: Language. Cambridge, MA: MIT Press.Pinker, S., Frost, L., & Lebeaux, D. (1987). Productivity and constraints in the acquisition of the passive. Cognition, 26, 195–267.Poeppel, D., & Wexler, K. (1993). The full competence hypothesis of clause structure in early German. Language, 69, 1–33.Radford, A. (1990). Syntactic theory and the acquisition of English syntax: The nature of early child grammars of English. Oxford: Basil Blackwell.Rizzi, L. (1994). Some notes on linguistic theory and language development: The case of root infinitives. Language Acquisition, 3, 371– 393.Saah, K., & Goodluck, H. (1995). Island effects in parsing and grammar: Evidence from Akan. Linguistic Review, 12, 381–409.Santelmann, L. (1995). The acquisition of verb second grammar in child Swedish. PhD dissertation, Cornell University.Tanenhaus, M., & Trueswell, J. (1995). Sentence comprehension. In J. Miller & P. Eimas (Eds.), Speech, language and communication: Handbook of perception and cognition (2nd ed.). San Diego, CA: Academic Press.Thornton, R., & Crain, S. (1994). Successful cyclic movement. In T. Hoekstra & B. Schwartz (Eds.), Language acquisition studies in generative grammar. Amsterdam/Philadelphia: John Benjamins.Thornton, R., & Wexler, K. (1999). VP-ellipsis, binding theory and interpretation in child grammar. Cambridge, MA: MIT Press.Treiman, R., Clifton, C., Jr., Meyer, A.,[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... & Wurm, L. (2003). Language comprehension and production. In A. Healy & R. Proctor (Eds.), Comprehensive handbook of psychology. Vol. 4: Experimental psychology. New York: John Wiley & Sons.Valian, V. (1999). Input and language acquisition. In W. Ritchie & T. Bhatia (Eds.), Handbook of child language acquisition. San Diego, CA: Academic Press.Valian, V., & Aubry, S. (2005). When opportunity knocks twice: two-year-olds’ repetition of sentence subjects. Journal of Child Language, 32, 617–641.Valian, V., & Eisenberg, S. (1996). The development of syntactic subjects in Portuguese- speaking children. Journal of Child Language, 23, 103–128.Weverink, M. (1989). The subject in relation to inflection in child language. MA thesis, University of Utrecht.Wexler, K. (1999). Maturation and growth of grammar. In W. Ritchie & T. Bhatia (Eds.), Handbook of child language acquisition. San Diego, CA: Academic Press.Wexler, K. (2004). Theory of phrasal development: Perfection in child grammar. Plato’s Problems: Papers on Language Acquisition, MIT Working Papers in Linguistics, 48, 159– 209.Wijnen, F. (1996). Temporal reference and eventivity in root infinitives. MIT Occasional Papers in Linguistics, 12, 1–25. 78. 4Domain-General Learning CapacitiesJenny R. Saffran and Erik D. Thiessen As far as acquisition of language is concerned, it seems clear that reinforcement, casual observation, and natural inquisitiveness (coupled with a strong tendency to imitate) are important factors, as is the remarkable capacity of the child to generalize, hypothesize, and “process information” in a variety of very special and apparently highly complex ways which we cannot yet describe or begin to understand, and which may be largely innate, or may develop through some sort of learning or through maturation of the nervous system. The manner in which such factors operate and interact in language acquisition is com- pletely unknown. It is clear that what is necessary in such a case is research, not dogmatic and perfectly arbitrary claims, based on analogies to that small part of the experimental literature in which one happens to be interested. Noam Chomsky (1959), A review of Skinner’s “Verbal Behavior”Language is arguably the most complex system acquired by humans. This fact, combinedwith the tender age at which language is typically learned, suggests that infants mustcome to the task of language acquisition already possessing the machinery required tomaster human language. What remains unknown is the nature of this machinery. Doinfants possess dedicated domain-specific learning mechanisms, evolved for languageacquisition? Or do infants take advantage of existing learning mechanisms that are notdomain-specific to discover the structure of human language? In this chapter, we willconsider the current state of the art in disentangling these views. While some progresshas been made since Chomsky’s (1959) quotation reprinted above, much still remainsunknown. It is important to note at the outset that the distinction between domain-specific anddomain-general learning mechanisms is orthogonal to the nature/nurture issue (e.g.,Peretz, in press). These two theoretical debates are often confounded; there is a tendencyto assume that innateness entails domain-[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... specific knowledge and/or learning mecha-nisms. However, all learning mechanisms presumably require innate structure, otherwisethere would be no way to get learning off the ground. For example, connectionist 79. Domain-General Learning Capacities 69networks – the paramount examples of domain-general learning devices – entail a greatdeal of “innate” structure, from input representations to learning rules to architecturalconstraints (e.g., Elman et al., 1996). Domain-general learning mechanisms can thus beinnate, and domain-specificity can be learned (witness evidence for localized brain areassubserving learned tasks such as reading and writing). Domain-specific learning mechanisms are traditionally invoked when learningphenomena are observed that are not seen in other domains. By contrast, domain-general learning mechanisms are invoked when parallel learning phenomena areobserved across distinct domains. Importantly, identical learning mechanisms can rendervery different kinds of knowledge in different domains. This is due to the fact that dif-ferent domains have different regularities, and infants face different constraints uponlearning in different domains. Because of this, a detailed look at the structureof the to- be-learned domain, along with a close investigation of the operation of anypotential learning mechanisms, is necessary before drawing conclusions about domain-generality or domain-specificity. To this end, this chapter will consider relevant empiricalevidence and evaluate the extent to which domain- general learning capacities canaccount for the acquisition of natural languages. In particular, we will focus on theareas of speech perception, speech category learning, word segmentation, word learning,and syntax, aspects of language where domain-specificity has been an explicit focus ofinvestigation.Historical Issues: Chomsky versus SkinnerThe conflict between domain-specific and domain-general views of language acquisitionhas its roots in an influential debate from the mid-twentieth century, with reverberationsthat extended far beyond the field of language. In 1957, B. F. Skinner published hisclassic volume, Verbal Behavior, which laid out his behaviorist theory of language acquisi-tion. Skinner invoked equipotential mechanisms for language acquisition via operantconditioning: the detection of contingencies between observable entities. Languageacquisition could thus be explained based on the organism’s history of experiences andreinforcement, via the same mechanisms observed for learning in other domains andspecies. In his devastating critique of Skinner’s theory, Chomsky (1959) argued convincinglythat internal representations are needed to explain language behavior. An internalizedgrammar allows learners to go beyond the particular sentences in the input, permittinggeneralization. By structuring the problem of language learning around the acquisitionof a grammar, Chomsky radically altered the field’s conceptualization of what languageacquisition entails. This, in turn, suggested a need for more specialized learning mecha- nisms: “The fact that all normal children acquire essentially comparable grammars ofgreat complexity with remarkable rapidity suggests that human beings are somehowspecially designed to do this, with data-handling or[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... ‘hypothesis-formulating’ ability ofunknown character and complexity” (Chomsky, 1959). Subsequent theoretical innova-tions led to a proposed language acquisition device – innate linguistic knowledge in the 80. 70 Jenny R. Saffran and Erik D. Thiessenform of a universal grammar, tied to dedicated language learning processes (Chomsky,1965, 1968). Chomsky’s early views continue to be extremely influential. In particular,there is no doubt that Skinner’s central claims were incorrect; external reinforcementcannot explain child language acquisition. However, recent research has begun toexamine other potentially general learning mechanisms that may play a role in languageacquisition; these theoretical and empirical innovations will be the focus of the remain-der of this chapter.Speech PerceptionSpeech is a uniquely human capacity that is closely tied to language. As such, speechperception is often regarded as a likely domain in which to find evidence for domain-specific learning mechanisms. Many aspects of this investigation can be viewed as attemptsto answer a deceptively simple question: Is speech special? That is, does speech perceptioninvoke unique (and uniquely human) processes? One of the most compelling argumentsadvanced in favor of the claim that speech is special is based upon the phenomenon ofcategorical perception in speech perception. Categorical perception is said to occur whendiscrimination is determined by category identification: listeners discriminate between-category contrasts, but cannot discriminate between members of the same category. Forexample, in Liberman, Harris, Hoffman, and Griffith’s (1957) classic experiment, listenerswere able to discriminate more easily between /b/ and /d/ (a cross- category distinction)than between two different examples of /b/, even though the two examples of /b/ wereas acoustically different as the cross-category pair. Other early experiments indicated thatdiscrimination of non-speech stimuli was continuous, not categorical (Mattingly, Liber-man, Syrdal, & Halwes, 1971), and that even very young infants show evidence of cate- gorical perception for speech (Eimas, Siqueland, Jusczyk, & Vigorito, 1971). Theseresults, and many others, were consistent with the theory that speech perception involvesunique processes not seen in other domains (e.g., Eimas, 1974). Later evidence was not consistent with this theory. Animal experiments demonstratedthat a variety of non-human species perceive speech sounds categorically (e.g., Kuhl &Miller, 1975). Further, both adults and infants show categorical perception for manynon-speech stimuli, including music-like sounds (e.g., Cutting & Rosner, 1974), faces(e.g., Etcoff & Magee, 1992), and color (e.g., Bornstein, Kessen, & Weiskopf, 1976).Categorical perception is more robust for stop consonants (e.g., /b/ and /k/) than it isfor vowels (e.g., Pisoni, 1975). These differences have led to proposals that there arespecialized memory systems for stop consonants and vowels (e.g., Schouten & vanHessen, 1992). However, categorical perception can also be observed to different degreesfor non-speech sounds that differ in the extent to which they are characterized by rapidlychanging acoustic dimensions – just[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... as the acoustic information associated with stopconsonants in speech changes more rapidly than vowels (Mirman, Holt, & McClelland,2004). Considering the wide variety of domains in which categorical perception can beobserved, recent theories and modeling work suggest that categorical perception maybe an inherent byproduct of perception in any domain where sufficiently dense stimuli 81. Domain-General Learning Capacities 71have a categorical structure, though of course categorical perception for speech is alsoinfluenced by the characteristics of the peripheral auditory system (e.g., Damper &Harnad, 2000). In addition to categorical perception, a number of other phenomena were initiallyconsidered to support claims of unique processing/learning for speech stimuli. Forexample, duplex perception occurs when speech sounds, split into two streams and pre-sented binaurally, simultaneously give rise to two distinct perceptual experiences (Fowler& Rosenblum, 1990). Initially, duplex perception was thought to occur only with speech;however, similar phenomena are seen with musical stimuli (e.g., Hall & Pastore, 1992).Similarly, the right ear advantage – in which sounds presented to the right ear can bedetected at lower amplitudes than sounds presented to the left ear – was initially linkedto the left hemisphere’s specialization for language (e.g., Glanville, Best, & Levenson,1977). However, the right ear advantage can also be demonstrated for non-linguisticstimuli such as tones or the “dot–dot–dashes” of Morse code in highly trained Morseoperators (Brown, Fitch, & Tallal, 1999; Papçun, Krashen, Terbeek, Remington, &Harshman, 1974). Finally, the McGurk effect – an effect of visual information on theperception of an auditory stimulus (MacDonald & McGurk, 1978) – can also be foundin the realm of music perception (Saldaña & Rosenblum, 1993). The fact that all ofthese phenomena also characterize non-linguistic perception suggests that the underlyingmechanisms are domain-general, not specific to language. While behavioral parallels between speech and non- speech domains are compelling,they fail to address an important question: Why are speech and language processingso consistently organized across individuals, tending to be centralized in the left hemi-sphere (e.g., Hickok, 2001)? Indeed, there are clear neurological differences betweenprocessing a sound when it is perceived as speech and when it is perceived as non- speech,even if the stimulus is identical (e.g., Dehaene-Lambertz et al., 2005). Further,event-related potential (ERP) data indicate that phoneme processing may invokesubstantially similar neurological processes early in infancy and in adulthood (Dehaene-Lambertz & Baillet, 1998; Dehaene-Lambertz & Gliga, 2004). Given the behavioralsimilarities between processing speech and non- speech (e.g., categorical perception,duplex perception), it seems initially incongruous that there would be a brain regiondedicated to speech processing, one that is at least partially consistent between infancyand adulthood. Note, however, that different brain regions need not imply different domain-specificlearning mechanisms. Zatorre and colleagues (Zatorre, 2001; Zatorre, Belin, & Penhune,2002) have suggested that the left hemisphere may[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... be better suited to processing tran-sient stimuli that require high temporal resolution due to the nature of its neural con-nections. Further, regional specificity in the brain may be related to factors other thanunique learning mechanisms. For example, many species preferentially attend to same- species vocalizations. This preference appears to be mediated by the activation of specificbrain regions that represent or respond to same-species utterances (Wang & Kadia, 2001;Wang, Merzenich, Beitel, & Schreiner, 1995). If, in these species, neural specializationor specific recruitment is related to attentional biases toward conspecific vocalizations,the same may be true of human infants. Infants’ brains may be geared to be particularlyresponsive to human speech. Consistent with this view, Vouloumanos and Werker 82. 72 Jenny R. Saffran and Erik D. Thiessen(2004) demonstrated that even 2-month-old infants have a reliable preference for humanspeech over a variety of other auditory stimuli. Therefore, it may be that speech is special in one important way: infants appear toattend preferentially to speech, which may ensure that speech is a particularly importantfeature in their environment (Vouloumanos & Werker, 2004). The phenomena thatappear unique to speech, or more frequently observed in speech, may in fact arise froman interaction between the acoustic characteristics of speech and our extensive experiencelistening to speech. This supposition is borne out by the fact that similar phenomenaarise in other domains with comparable stimulus density (e.g., categorical perception inface recognition), acoustic characteristics (e.g., right ear advantage for tone discrimina-tion, duplex perception of chords), and familiarity (e.g., right ear advantage in trainedMorse code operators.). While early-developing or innate attentional biases favoringspeech ensure that it is highly salient in a way few other stimuli are, there is mountingevidence to suggest that speech perception is influenced by the same learning mecha- nisms that are responsible for processing other types of stimuli. This parallel is mostclearly seen in face perception: infants’ early propensity to attend to face-like stimulicombines with subsequent experience to affect children’s categorical perception of facialemotion displays (Pollak & Kistler, 2002).Speech CategoriesAcross different languages, different acoustic contrasts are meaningful. For example, thedistinction between /r/ and /l/ indicates different meanings in English (the differenceis “phonemic,” as in “rock” vs. “lock”), but not in Japanese. Infants must learn whichacoustic distinctions are productive in their linguistic environment. This knowledge isacquired rapidly; infants adapt their responses to the phonemic categories of their lan-guage within the first year of life (see Polka, Rvachew, & Mattock, this volume). Themechanism that makes this learning possible may involve sensitivity to the statisticalstructure of the linguistic input, in the form of the distribution of speech sounds in thelinguistic environment (see Gerken, this volume). A variety of non-human animals showsimilar attunement to particular acoustic contrasts in response to information about thedistribution of speech sounds (e.g., Kluender, Lotto, Holt, & Bloedel, 1998). This sug-gests that[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... learning about phonemic categories may arise from the same mechanisms aslearning about any type of category. Even newborns can categorize (Slater, 1995). Just as infants adapt to the speech soundcategories of their native language in response to perceptual information, infants’ earlyobject categories are based on perceptual, rather than conceptual, information (e.g.,Mandler, 2000; Quinn & Eimas, 2000). Also, just as distributional information playsan important role in infants’ adaptation to phonemic categories, the frequency and dis-tribution of infants’ experience with different exemplars influences developing objectcategories. When presented with a set of exemplars with a highly variable distribution,infants form broad, inclusive categories. When familiarized with a more focused distri-bution of exemplars, infants form categories with tighter boundaries (e.g., Oakes & 83. Domain-General Learning Capacities 73Spalding, 1997). In a similar vein, Huttenlocher, Hedges, and Vevea (2000) demon-strated that adults’ identification of exemplars is influenced by the previous distributionof category members they have experienced. Huttenlocher, Hedges, Corrigan, and Crawford (2004) argue that a process criticalto inductive categorization is the formation of categories that capture the distributionaldensity of previously experienced exemplars. Ideally, categories should be formed with aprototypical member near the center of the distribution, the region of the highest densityof exemplars. Category boundaries should be placed in regions with low exemplardensity. Categories with these characteristics are efficient; placing category boundariesin sparsely populated regions means that there is less likelihood of misclassifying stimuli.Maye, Werker, and Gerken (2002) demonstrated that adults and infants place stimulusboundaries in regions of low density in response to different distributions of speechsounds. Therefore, it seems quite plausible that infants’ ability to adapt to their language’scategories of speech sounds may be a specific instance of a more general tendency to usedistributional information as a cue to categorization. Note, however, that previous experi-ence can prevent learners from forming efficient categories. Most famously, previousexperience with a language that does not use a phonemic contrast (such as /r/ and /l/ inJapanese) can lead to difficulty acquiring the distinction in response to new acousticdistributions in a new linguistic environment. Similarly, perceptual biases can influencecategory formation. Japanese listeners find English /r/ to be more dissimilar to Japanese/r/ than English /l/, which has been proposed to explain why native Japanese speakersshow more improvement in their use of /r/ than /l/ when learning English (Aoyama,Flege, Guion, Akahane- Yamada, & Yamada, 2004). Infants’ and adults’ discovery ofspeech categories is likely to be strongly influenced by the similarity of different speechsounds in their language. Infants need not only to learn which acoustic distinctions are phonemic in their lan-guage. They must also learn how to appropriately produce the sounds comprising thephonemic inventory of their language. Perceptual input will, of course, play an impor-tant role in specifying infants’ productive repertoire. Additional learning mechanismsmust play a role,[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... though. Goldstein, King, and West (2003) have demonstrated thatsocial shaping plays an important role in allowing infants to converge upon language-appropriate verbal behavior, paralleling the development of birdsong. Sounds that receivemore social response are more likely to recur, shaping the communicative inventory forfuture interactions, as demonstrated in the domain of infant babbling (Goldstein et al.,2003). Thus, humans and non-humans may share some learning mechanisms thatsupport the development of productive communicative abilities.Word SegmentationUnlike the blank spaces between words in text, speakers do not consistently place pausesbetween words in fluent speech. This presents a challenge to young infants who mustlocate word boundaries. Despite the complexity of this task, infants are able to segmentwords from fluent speech by at least 7 months of age (Jusczyk & Aslin, 1995). 84. 74 Jenny R. Saffran and Erik D. Thiessen One cue that allows infants to discover words in fluent speech is sequential statisticalinformation. Syllables within a word are more likely to occur together than syllables thatare not part of the same word. Saffran, Aslin, and Newport (1996) provided evidence thatboth infants and adults are capable of using transitional probabilities between syllables todetect word boundaries in fluent speech. Statistical learning mechanisms are availableacross species (e.g., Hauser, Newport, & Aslin, 2001), and in a variety of domains. Adultsand infants attend to transitional probabilities in visual stimuli and non-linguistic auditorystimuli (Fiser & Aslin, 2001; Saffran, Johnson, Aslin, & Newport, 1999). Sequential statistical cues are available to infants from very early in life (Kirkham,Slemmer, & Johnson, 2002), and may play a role in infants’ earliest segmentation ofwords from fluent speech (Thiessen & Saffran, 2003). However, infants also use anotherkind of cue to word segmentation: acoustic cues. For example, in English, stress is cor-related with word beginnings, and between 8 and 9 months, infants begin to treatstressed syllables as word onsets (Jusczyk, Houston, & Newsome, 1999). While younginfants favor transitional probabilities over stress cues, older infants rely more on stresscues (Johnson & Jusczyk, 2001). Infants may learn to use these acoustic cues to wordboundaries via the same statistical learning abilities that allow infants to take advantageof transitional probabilities. Statistical learning can be more broadly construed as attention to regularities in theenvironment. Attending to such regularities allows learners to discover which eventspredict other events (e.g., Canfield & Haith, 1991). On this interpretation, attention totransitional probabilities between elements in sequence is only one particular exampleof statistical learning. To discover acoustic regularities such as lexical stress, infantsrequire experience to indicate which acoustic events have predicted word positions onprevious occasions. To do so, infants must be familiar with at least a few words (possiblydiscovered via transitional probabilities in fluent speech, or heard in isolation). Fromthese words, infants can detect which acoustic characteristics are correlated with wordpositions; for example, once infants are familiar with a few words, it is possible for[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... themto discover that most of those words begin with a stressed syllable, and to begin to treatstress as a cue to word onsets (Swingley, 2005). Chambers, Onishi, and Fisher (2003)have suggested that similar learning mechanisms may allow infants to discover whichsound combinations are permissible in their language. For example, in English, “fs” isnot permitted in word-initial position; discovering these types of regularities can helpinfants segment fluent speech (Mattys, Jusczyk, Luce, & Morgan, 1999). With age andexperience, infants become able to integrate multiple cues to word segmentation (e.g.,Morgan & Saffran, 1995). A similar developmental progression – from reliance on singlecues to weighting of multiple cues – is seen in object categorization (Younger & Cohen,1986). Therefore, it seems likely that the developmental trajectory underlying infants’use and integration of multiple cues arises from domain-general processes.Words and MeaningLearning the meaning of words is one of the great milestones of early development. Themajority of the research on infants’ word learning has focused on nouns. In this context, 85. Domain-General Learning Capacities 75“meaning” refers to the connection between a noun and the object to which it refers.This connection is often assessed via comprehension measures (such as looking), becauseinfants comprehend far more words than they can produce (e.g., Benedict, 1979). Wordlearning is slow before the first birthday, but it does occur; for example, 6-month-oldslook longer at a picture of their mother in response to the word “Mommy” (Tincoff &Jusczyk, 1999). Between their first and second birthday, children begin to learn wordsmore easily (Bloom, 2000; Werker, Cohen, Lloyd, Casasola, & Stager, 1998). One ofthe most impressive abilities children demonstrate during this period is “fast-mapping,”the ability to form a connection between words and referents with as little as one expo- sure (Heibeck & Markman, 1987). This seemingly unique phenomenon has promptedspeculation that humans may possess a dedicated word-learning mechanism (e.g.,Waxman & Booth, 2000). To assess the claim that word learning is the result of a domain-specific mechanism,we must examine the processes that enable word learning. One process that is critical toword learning is the ability to detect correspondences between words and objects, andto form an association between them. Objects that are regularly present when a wordoccurs are likely candidates as referents for that word, at least for nouns (e.g., Plunkett& Schafer, 1999; Roy & Pentland, 2002). However, word learning is also influenced bya variety of adaptive biases and constraints. The shape bias, for example, refers to chil-dren’s tendency to generalize names to novel objects on the basis of shape (Landau,Smith, & Jones, 1988). The principle of mutual exclusivity holds that any object hasonly one label (Markman & Wachtel, 1988). The whole object bias refers to children’spreference to treat labels as referring to whole objects, rather than parts of objects (Soja,Carey, & Spelke, 1991). Finally, when children learn the name of an object, they tendto treat that label as a reference to a class of objects (such as dogs), rather than a singleobject; this is called the taxonomic[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... bias (Markman & Hutchinson, 1984). The origin of these biases is uncertain. One possibility is that they are both innateand specific to language, as has been argued for the whole object, taxonomic, and mutualexclusivity biases (e.g., Markman, 1991, but see Markman, 1992; for discussion, seeDiesendruck, this volume). Alternatively, these biases could be specific to language, butarise from domain-general learning mechanisms. Infants’ early experience with wordsmay highlight linguistic regularities that facilitate subsequent word learning (e.g.,Landau, Gershkoff-Stowe, & Samuelson, 2002; Samuelson, 2002). Finally, it may bethe case that these biases are the result of domain-general constraints on the mechanismsthat make word learning possible. For example, mutual exclusivity may arise from thefact that forming an association between two stimuli, X and Y, makes forming subse- quent associations between one of those stimuli (X) and a new stimulus (Z) more diffi-cult (e.g., Mackintosh, 1971). Many questions remain to be answered about the parallelsbetween word learning and learning in other domains before it will be clear which ofthese positions is correct (e.g., Halberda, 2003; Sabbagh & Gelman, 2000). A third source of information for word learning is social interaction. Infants are sensi-tive to the social intent of speakers in word-learning situations (see Baldwin & Meyer,this volume). An important direction for future research will be to examine the interac-tions between these sources of information. For example, there are likely to be severalpotential referents in the environment each time a word occurs. If an infant depends onstatistical information alone, word learning will be extremely difficult (Bloom, 2000). 86. 76 Jenny R. Saffran and Erik D. ThiessenSocial cues to referential intent can facilitate word learning. Similarly, the effect of con-straints on word learning can be influenced by the pragmatic and perceptual context inwhich words are taught (Diesendruck, Gelman, & Lebowitz, 1998). As these examplesillustrate, multiple domain-general learning mechanisms – such as statistical learningand social learning – can combine to create domain-specific knowledge (the meaningof words). Any domain-general account of word learning, though, must account for fast-mapping,the signature phenomenon thought to demonstrate a unique mechanism for wordlearning. Critically, Markson and Bloom (1997) have demonstrated that children“fast-map” novel facts about objects as well as their names. These results suggest thatfast- mapping is a specific realization of a general capacity. Evidence for fast- mapping ina dog (Kaminski, Call, & Fischer, 2004) has similar implications. However, the possibil-ity that lexical learning arises from domain-general mechanisms does not imply thatlexical learning proceeds in precisely the same manner as other types of learning. Forexample, whereas children extend names to other objects in the same category, childrenare more limited in their extensions of facts (e.g., Waxman & Booth, 2000). This dif-ference may be due to the fact that general learning mechanisms render different knowl-edge as a function of the structure of the domain being acquired (Saffran, 2001a). Whenchildren learn the names of objects, those names[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... frequently apply to all of the otherobjects in that category. Facts (such as “my uncle gave me this one”) apply only to oneindividual object; they are like proper nouns (Bloom & Markson, 2001). The contrastbetween facts and words illustrates that learning mechanisms can give rise to very dif-ferent knowledge based on children’s experience.SyntaxAlong with speech perception, syntax is the aspect of language where domain-specificityhas been most widely assumed. This domain-specificity takes two forms: innate linguis-tic knowledge (in the form of a universal grammar) and domain- specific learning mecha-nisms (e.g., triggering mechanisms in the principles and parameters framework; seeGoodluck, this volume). Domain-specificity has been implicit in many of these theoriesfor at least three reasons. First, syntax is typically abstract, and not transparently mir-rored in the surface structure of the input, suggesting the need for dedicated machinery.Second, the languages of the world contain remarkably little syntactic variation, a factthat is readily explained by hypothesizing innate linguistic knowledge (e.g., Baker, 2001).Third, non-human animals have difficulty acquiring human syntactic structures; thesespecies differences can be explained by hypothesizing dedicated human linguisticmachinery. However, evidence is mounting that at least some syntactic regularities may be learn-able by domain-general mechanisms. For example, consider the acquisition of grammati-cal categories – determining which words are nouns, which are verbs, etc. Children areable to appropriately use grammatical category information by the middle of the secondyear (e.g., Bloom, 1970; Brown, 1973). Prominent semantic bootstrapping accounts of 87. Domain-General Learning Capacities 77this phenomenon rely on innate linguistic knowledge concerning semantic–syntacticcorrespondences (e.g., Pinker, 1984). More recent accounts, however, building from anearlier proposal by Maratsos and Chalkley (1980), have argued that infants coulddiscover which words cohere into grammatical categories by tracking patterns of co-occurrence of words in the input (e.g., Mintz, Newport, & Bever, 2002; Redington,Chater, & Finch, 1998). For example, one might discover the category Noun by deter-mining that a certain set of words was typically preceded by “the.” While there are manyindividual counterexamples (Pinker, 1985), computational analyses suggest that theinformation needed to cluster words into categories is available in child-directed speech,and adults learning artificial languages can discover grammatical categories using solelydistributional information (Mintz, 2002). While these findings do not directly demon-strate domain-generality, as the materials are always linguistic, categorization via distri-butional information is unlikely to be limited to language learning. Other lines of research have directly addressed the issue of domain-generalversus domain-specific learning mechanisms by contrasting the use of linguistic andnon-linguistic “grammars.” Building on research by Morgan and Newport (1981;Morgan, Meier, & Newport, 1987, 1989), Saffran (2001b, 2002) investigated the use ofdistributional information for discovering linguistic phrase structure, a widespreadaspect of syntactic[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... structure cross-linguistically. Adults can use a statistical cue to phrasalunits, predictive dependencies (e.g., the presence of “the” or “a” strongly predicts a nounsomewhere downstream), to discover phrase boundaries (Saffran, 2001b). Moreover,adults and children are better at acquiring languages that contain predictive dependen-cies than those that do not (Saffran, 2002). Interestingly, the same constraint on learningemerges in tasks using non- linguistic materials, including both auditory non-linguisticgrammars (in which the “words” were computer alert sounds) and visual non-linguisticgrammars (simultaneously presented arrays of shapes). Saffran (2002, 2003a) hypothe- sized that this domain-general learning mechanism has played a role in shaping thestructure of natural languages. On this view, languages contain predictive dependenciesas cues to phrasal units because this information helps human learners to discover phrasesin natural languages. A domain- general learning ability may have shaped the structureof something quite specific – language.Challenges for Domain-General AccountsHow does one “prove” that a learning mechanism is domain-general? Even the clearestcases – where learners show equivalent performance when acquiring materials from twodifferent domains, given the same patterns in the input – could equally well representtwo parallel learning mechanisms in lieu of a single domain-general mechanism. Herewe see a logical problem with demonstrations of domain-generality: while parsimonymight suggest that one learning mechanism is better than two, the natural world is notalways parsimonious. One approach to this problem would be to attempt to identify theneural basis of the learning mechanisms in question. In this attempt, though, it isimportant to remember that the use of distinct brain areas by expert users of a system 88. 78 Jenny R. Saffran and Erik D. Thiessen(e.g., adults) does not necessarily signal the use of distinct learning mechanisms (fordiscussion, see McMullen & Saffran, 2004; Peretz, in press). Other objections to claims of domain-generality arise from the empirical data them-selves. Some mechanisms used for language learning, such as rule-pattern detection(Marcus, Vijayan, Rao, & Vishton, 1999), may not readily operate over all non-linguisticstimuli (Marcus, Johnson, & Fernandes, 2004). However, they are apparently usable bynon-human primates (Hauser, Weiss, & Marcus, 2002), and do operate over at leastsome non-linguistic stimuli (Saffran, Pollak, Seibel, & Shkolnik, in press). More gener-ally, debate continues over the degree to which complex syntactic structures requiredomain-specific innate knowledge, or whether they can instead be explained with refer-ence to more general cognitive/social/pragmatic mechanisms (see Lidz, this volume).Evidence from circumstances in which children create their own languages, as in creoliza-tion (e.g., Senghas, Kita, & Ozyurek, 2004) and homesign (e.g., Goldin-Meadow, 2003),may help to resolve some of these issues. In such cases, there is a far greater divergencebetween the structure of the input and the child’s eventual linguistic attainments, allow-ing for a careful parsing of the types of learning mechanisms in operation. Other[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... objections stem from the overall contour of the evidence concerning childlanguage acquisition. For example, if children are relying on general learning mecha-nisms to acquire language, then why are they markedly more successful than non-humanprimates? That is, if language learning doesn’t rely on anything special about language,why do only humans do it so well? The answer to such objections may well lie in thespecifics of how the learning mechanisms work – as opposed to taking “language” and“cognition” as unitary constructs. For example, there are likely to be specific cognitivedifferences between humans and non-humans that may affect language learning, evenif these differences did not evolve specifically to support language acquisition (e.g.,Hauser, Chomsky, & Fitch, 2002). Recent evidence points to differences in the use oflearning mechanisms across species that may affect language learning outcomes. Forexample, given transitional probabilities computed over non-adjacent syllables (withother syllables intervening between the target syllables), human and tamarin learnersshow quite different patterns of performance (Newport & Aslin, 2004; Newport, Hauser,Spaepen, & Aslin, 2004). Both species show limitations in the types of patterns theydetect. Critically, however, the kinds of limitations observed in humans map onto naturallanguage structures – segmental non- adjacency patterns that occur in languages arelearnable by humans – whereas the tamarins’ learning abilities appear to be unrelatedto the structures observed in natural languages. The fact that humans also exhibit relatedconstraints when acquiring non-linguistic sequences such as tones (Creel, Newport, &Aslin, 2004) supports the contention that non-linguistic limitations on what is learnablemay have shaped the organization of human languages. A related objection pertains to the contrast between child and adult learners. If lan-guage acquisition rests on general learning abilities, then wouldn’t one expect adults tooutperform children, when in fact the available evidence suggests that it is the other wayaround? Again, this sort of objection makes the assumption that there is some sort ofoverarching “general learning ability.” This apparent paradox may be resolved by con-sidering other features of cognition that distinguish children and adults. For example,Newport (1990) has argued that children’s relatively constrained working memory 89. Domain-General Learning Capacities 79capacities may in fact facilitate some aspects of language learning. Combinatorial systemslike morphology and syntax require the discovery of small component pieces of languagein order to discover the patterns that relate them. The sieve-like nature of children’smemories might facilitate the discovery of these pieces, whereas adults are more likelyto remember larger chunks of language, missing the underlying patterns. Consistent withthis hypothesis, adults actually appear to learn certain aspects of novel languages moresuccessfully when engaged in a concurrent capacity-limiting task (Cochran, McDonald,& Parault, 1999). Evidence from atypical development is often raised in objections to domain- generalaccounts. The classic picture is that of a double dissociation, in which[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... “language” isspared while “cognition” is disrupted as in Williams syndrome (WS) (e.g., Pinker, 1991;Rossen, Jones, Wang, & Klima, 1995), while the opposite pattern is obtained in specificlanguage impairment (SLI) (Crago & Gopnik, 1994; Rice, 1999). These kinds of find-ings are taken as evidence for a distinction between abilities used to learn language andthe rest of cognition; for example, Pinker (1999) contrasts individuals with SLI and WSby noting that “the genes of one group of children impair their grammar while sparingtheir intelligence; the genes of another group of children impair their intelligence whilesparing their grammar” (p. 262). One reason that this picture of a clean double dissocia-tion originally emerged is that language and cognition were each taken as unitary con-structs. However, when the multiple interlocking subcomponents of language andcognition are considered, the picture of strengths and weaknesses within particularpopulations becomes more complex (e.g., Shatz, 1994). For example, individuals withWS show atypical language abilities in a number of subdomains, from word segmenta- tion (Nazzi, Paterson, & Karmiloff-Smith, 2003) to morphosyntax (Karmiloff- Smithet al., 1997), suggesting that the intact language hypothesis in this population is a myth(for review, see Karmiloff-Smith, Brown, Grice, & Paterson, 2003). Similarly, individu-als with SLI show impairments in non- “core” language abilities such as speech perception(Joanisse & Seidenberg, 1998), the use of symbolic representation (Johnston & Ramstad,1983), and verbal working memory (Weismer, Evans, & Hesketh, 1999). This morecomplex picture of these disorders does not rule out the existence of specialized learningcapacities. However, it does suggest that the classic double-dissociation argument is lessclearly applicable than was previously believed.Conclusions“Domain-general” is a loaded term. It implies a set of generalized simple learning devicesthat can operate over any types of input, such as those espoused by Skinner. The litera-ture that we have reviewed suggests that this is an overly simplistic view of the learningabilities that likely contribute to language learning. These learning mechanisms areconstrained to operate over some types of input but not others, as a function of humanperception and cognition. They may incorporate both innate and emergent properties.And much of the power of the mechanisms in question likely lies in the ways in whichthey mutually interact; for example, once learners perform distributional analyses that 90. 80 Jenny R. Saffran and Erik D. Thiessenrender categories, the input to learning changes, such that learners can begin to acquirepatterns over categories (types) rather than over the raw input (tokens). “Domain-specific” is also a loaded term, which usually implies an innate, modular,knowledge system. It is evident, however, that domain-specificity and innateness arerightly viewed as orthogonal variables. Modularity can emerge as a function of experiencewithin a particular domain. While the adult state clearly involves some localization ofcognitive and linguistic functions, this domain- specificity might be the end-result ofdomain-general mechanisms operating on material drawn from different input domains(e.g., McMullen & Saffran,[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... 2004). The structure of the input-to-be-learned will influ-ence the eventual outcome of learning, such that the same mechanism can obtain dif-ferent results as a function of prior knowledge about the input domain (e.g., Saffran,2001a, 2003b), the age of the learner (e.g., Saffran & Griepentrog, 2001), the structureof the input (Gerken, 2004; Saffran, Reeck, Niehbur, & Wilson, 2005), or the speciesof the learner (Newport et al., 2004). In addition, a developmental perspective is likelyto be quite useful in disentangling initial states from eventual outcomes, for both typi-cally and atypically developing populations (e.g., Karmiloff-Smith, 1998). Technologicaladvances may also facilitate researchers’ ability to ask whether distinct brain areas sub-serve the acquisition of distinct domains of knowledge early in infancy (e.g., Peña et al.,2003). Returning to the Chomsky (1959) quotation with which this chapter began, itis clear that continued research, rather than dogma, is needed in order to render themost significant progress on the question of domain- specificity and domain-generalityin language acquisition.NotePreparation of this chapter was supported by grants to the first author from NICHD(R01HD37466) and NSF (BCS-9983630). We thank Katharine Graf Estes, Michael Kaschak,Erin McMullen, Isabelle Peretz, and David Rakison for helpful comments on a previous draft.ReferencesAoyama, K., Flege, J. E., Guion, S. G., Akahane-Yamada, R., & Yamada, T. (2004). Perceived phonetic dissimilarity and L2 speech learning. Journal of Phonetics, 32, 233– 250.Baker, M. C. (2001). The atoms of language. New York: Basic Books.Benedict, H. (1979). Early lexical development: Comprehension and production. Journal of Child Language, 6, 183–199.Bloom, L. (1970). Language development: Form and function in emerging grammars. Cambridge, MA: MIT Press.Bloom, P. (2000). How children learn the meanings of words. Cambridge, MA: MIT Press.Bloom, P., & Markson, L. (2001). Are there principles that apply only to the acquisition of words? Cognition, 78, 89–90.Bornstein, M. H., Kessen, W., & Weiskopf, S. (1976). The categories of hue in infancy. Science, 191, 201–202. 91. Domain-General Learning Capacities 81Brown, C. P., Fitch, H. R., & Tallal, P. (1999). Sex and hemispheric differences for rapid audi- tory processing in normal adults. Lateralities: Asymmetries of Body, Brain, and Cognition, 4, 39–50.Brown, R. (1973). A first language: The early stages. London: George Allen & Unwin.Canfield, R. L., & Haith, M. M. (1991). Young infants’ visual expectations for symmetric and asymmetric stimulus sequences. Developmental Psychology, 27, 198–208.Chambers, K. E., Onishi, K. H., & Fisher, C. (2003). Infants learn phonotactic regularities from brief auditory experiences. Cognition, 87, B69–B77.Chomsky, N. (1959). A review of B.F. Skinner’s “Verbal Behavior.” Language, 35, 26–58.Chomsky, N. (1965). Aspects of the theory of syntax. Cambridge, MA: MIT Press.Chomsky, N. (1968). Language and mind. New York: Harcourt Brace Jovanovich.Cochran, B. P., McDonald, J. L., & Parault, S. J. (1999). Too smart for their own good: The disadvantage of a superior processing capacity for adult language learners. Journal of Memory and Language, 41, 30–[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... 58.Crago, M. B., & Gopnik, M. (1994). From families to phenotypes: Theoretical and clinical implications of research into the genetic basis of specific language impairment. In R. Watkins & M. Rice (Eds.), Specific language impairments in children (pp. 35–51). Baltimore, MD: Brookes.Creel, S. C., Newport, E. L., & Aslin, R. N. (2004). Distant melodies: Statistical learning of non-adjacent dependencies in tone sequences. Journal of Experimental Psychology: Learning, Memory, and Cognition, 30, 1119– 1130.Cutting, J. E., & Rosner, B. S. (1974). Categories and boundaries in speech and music. Perception and Psychophysics, 16, 564–570.Damper, R. I., & Harnad, S. R. (2000). Neural network models of categorical perception. Perception and Psychophysics, 62, 843–867.Dehaene-Lambertz, G., & Baillet, S. (1998). A phonological representation in the infant brain. NeuroReport, 9, 1885–1888.Dehaene-Lambertz, G., & Gliga, T. (2004). Common neural basis for phoneme processing in infants and adults. Journal of Cognitive Neuroscience, 16, 1375–1387.Dehaene-Lambertz, G., Pallier, C., Serniclaes, W., Sprenger-Charolles, L., Jobert, A., & Dehaene, S. (2005). Neural correlates of switching from auditory to speech perception. NeuroImage, 24, 21–33.Diesendruck, G., Gelman, S. A., & Lebowitz, K. (1998). Conceptual and linguistic biases in children’s word learning. Developmental Psychology, 34, 823–839.Eimas, P. D. (1974). Auditory and linguistic processing of cues for place of articulation by infants. Perception and Psychophysics, 16, 513–521.Eimas, P. D., Siqueland, E. R., Jusczyk, P. W., & Vigorito, J. (1971). Speech perception in infants. Science, 171, 303– 306.Elman, J. L., Bates, E. A., Johnson, M. H., Karmiloff-Smith, A., Parisi, D., & Plunkett, K. (1996). Rethinking innateness: A connectionist perspective on development. Cambridge, MA: MIT Press.Etcoff, N. L., & Magee, J. J. (1992). Categorical perception of facial expressions. Cognition, 44, 227–240.Fiser, J., & Aslin, R. N. (2001). Unsupervised statistical learning of higher-order spatial structures from visual scenes. Psychological Science, 12, 499–504.Fowler, C. A., & Rosenblum, L. D. (1990). Duplex perception: A comparison of monosyllables and slamming doors. Journal of Experimental Psychology: Human Perception and Performance, 16, 742–754. 92. 82 Jenny R. Saffran and Erik D. ThiessenGerken, L. (2004, May). Infants make different generalizations for the same formal system depending on the specifics of the input. Paper presented at the 14th International Conference on Infant Studies, Chicago.Glanville, B. B., Best, C. T., & Levenson, R. (1977). A cardiac measure of cerebral asymmetries in infant auditory speech perception. Developmental Psychology, 13, 54–59.Goldin-Meadow, S. (2003). The resilience of language: What gesture creation in deaf children can tell us about how all children learn language. New York: Psychology Press.Goldstein, M. H., King, A. P., & West, M. J. (2003). Social interaction shapes babbling: Testing parallels between birdsong and speech. Proceedings of the National Academy of Sciences of the USA, 100, 8030– 8035.Halberda, J. (2003). The development of a word-learning strategy.[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Cognition, 87, B23–B34.Hall, M. D., & Pastore, R. E. (1992). Musical duplex perception: Perception of figurally good chords with subliminal distinguishing tones. Journal of Experimental Psychology: Human Per- ception and Performance, 18, 752–762.Hauser, M. D., Chomsky, N., & Fitch, W. T. (2002). The faculty of language: What is it, who has it, and how did it evolve? Science, 298, 1569–1579.Hauser, M. D., Newport, E. L., & Aslin, R. N. (2001). Segmentation of the speech stream in a non-human primate: Statistical learning in cotton-top tamarins. Cognition, 78, B53– B64.Hauser, M. D., Weiss, D., & Marcus, G. (2002). Rule learning by cotton-top tamarins. Cogni- tion, 86, B15–B22.Heibeck, T. H., & Markman, E. M. (1987). Word learning in children: An examination of fast mapping. Child Development, 58, 1021–1034.Hickok, G. (2001). Functional anatomy of speech perception and speech production: Psycholin- guistic implications. Journal of Psycholinguistic Research, 30, 225–235.Huttenlocher, J., Hedges, L. V., Corrigan, B., & Crawford, L. E. (2004). Spatial categories and the estimation of location. Cognition, 93, 75–97.Huttenlocher, J., Hedges, L. V., & Vevea, J. L. (2000). Why do categories affect stimulus judg- ment? Journal of Experimental Psychology: General, 129, 220–241.Joanisse, M., & Seidenberg, M. S. (1998). Specific language impairment: A deficit in grammar or processing? Trends in Cognitive Sciences, 2, 240–247.Johnson, E. K., & Jusczyk, P. W. (2001). Word segmentation by 8-month-olds: When speech cues count more than statistics. Journal of Memory and Language, 44, 548–567.Johnston, J. R., & Ramstad, V. (1983). Cognitive development in pre-adolescent language impaired children. British Journal of Disorders of Communication, 18, 49– 55.Jusczyk, P. W., & Aslin, R. N. (1995). Infants’ detection of the sound patterns of words in fluent speech. Cognitive Psychology, 29, 1–23.Jusczyk, P. W., Houston, D. M., & Newsome, M. (1999). The beginnings of word segmentation in English-learning infants. Cognitive Psychology, 39, 159– 207.Kaminski, J., Call, J., & Fischer, J. (2004). Word learning in a domestic dog: Evidence for fast mapping. Science, 304, 1682–1683.Karmiloff-Smith, A. (1998). Development itself is the key to understanding developmental dis- orders. Trends in Cognitive Sciences, 2, 389–398.Karmiloff-Smith, A., Brown, J. H., Grice, S., & Paterson, S. (2003). Dethroning the myth: Cognitive dissociations and innate modularity in Williams syndrome. Developmental Neuro- psychology, 23, 229–244.Karmiloff-Smith, A., Grant, J., Berthoud, I., Davies, M., Howlin, P., & Udwin, O. (1997) Language and Williams Syndrome: How intact is “Intact”? Child Development, 68, 246– 262. 93. Domain-General Learning Capacities 83Kirkham, N. Z., Slemmer, J. A., & Johnson, S. P. (2002). Visual statistical learning in infancy: Evidence for a domain general learning mechanism. Cognition, 83, B35–B42.Kluender, K. R., Lotto, A. J., Holt, L. L., & Bloedel, S. L. (1998). Role of experience for lan- guage-specific functional mappings of vowel sounds. Journal of the Acoustical Society of America, 104, 3568–3582.Kuhl, P. K., & Miller, J. D. (1975). Speech perception by the chinchilla: Voiced–voiceless dis- tinction in alveolar plosive consonants. Science, 190, 69–72.Landau, B., Smith, L. B., &[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Jones, S. S. (1988). The importance of shape in early lexical learn- ing. Cognitive Development, 3, 299–321.Liberman, A. M., Harris, K. S., Hoffman, H., & Griffith, B. (1957). The discrimination of speech sounds within and across phoneme boundaries. Journal of Experimental Psychology, 54, 358– 368.MacDonald, J., & McGurk, H. (1978). Visual influences on speech perception processes. Percep- tion and Psychophysics, 24, 253– 257.Mackintosh, N. J. (1971). An analysis of blocking and overshadowing. Quarterly Journal of Experimental Psychology, 23, 118–125.Mandler, J. M. (2000). Perceptual and conceptual processes in infancy. Journal of Cognition and Development, 1, 3–36.Maratsos, M., & Chalkley, M. A. (1980). The internal language of children’s syntax: The ontogenesis and representation of syntactic categories. In K. Nelson (Ed.), Children’s language: Vol. 2 (pp. 127– 214). New York: Gardner Press.Marcus, G. F., Johnson, S. P., & Fernandes, K. (2004, November). What’s special about speech? Evidence from a contrast between rules and statistics. Paper presented at the 29th Annual Boston University Conference on Language Development.Marcus, G. F., Vijayan, S., Rao, S. B., & Vishton, P. M. (1999). Rule learning by seven-month- old infants. Science, 283, 77–80.Markman, E. M. (1991). The whole-object, taxonomic, and mutual exclusivity assumptions as initial constraints on word meaning. In S. Gelman & J. Byrnes (Eds.), Perspectives on language and thought: Interrelations in development (pp. 72–106). Cambridge: Cambridge University Press.Markman, E. M. (1992). Constraints on word learning: Speculations about their nature, origins, and domain specificity. In M. R. Gunnar & M. Maratsos (Eds.), Modularity and constraints in language and cognition (pp. 59–101). Hillsdale, NJ: Erlbaum.Markman, E. M., & Hutchinson, J. E. (1984). Children’s sensitivity to constraints on word meaning: Taxonomic versus thematic relations. Cognitive Psychology, 16, 1–27.Markman, E. M., & Wachtel, G. F. (1988). Children’s use of mutual exclusivity to constrain the meaning of words. Cognitive Psychology, 20, 121–157.Markson, L., & Bloom, P. (1997). Evidence against a dedicated system for word learning in children. Nature, 385, 813–815.Mattingly, I. G., Liberman, A. M., Syrdal, A. K., & Halwes, T. (1971). Discrimination in speech and nonspeech modes. Cognitive Psychology, 2, 131–157.Mattys, S. L., Jusczyk, P. W., Luce, P. A., & Morgan, J. L. (1999). Phonotactic and prosodic effects on word segmentation in infants. Cognitive Psychology, 38, 465–494.Maye, J., Werker, J. F., & Gerken, L. (2002). Infant sensitivity to distributional information can affect phonetic discrimination. Cognition, 82, B101–B111.McMullen, E., & Saffran, J. R. (2004). Music and language: A developmental comparison. Music Perception, 21, 289–311.Mintz, T. H. (2002). Category induction from distributional cues in an artificial language. Memory and Cognition, 30, 678– 686. 94. 84 Jenny R. Saffran and Erik D. ThiessenMintz, T. H., Newport, E. L., & Bever, T. G. (2002). The distributional structure of grammati- cal categories in speech to young children. Cognitive Science, 26, 393–424.Mirman, D., Holt, L. L., & McClelland, J. L. (2004). Categorization and discrimination of[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... nonspeech sounds: Differences between steady-state and rapidly-changing acoustic cues. Journal of the Acoustical Society of America, 116, 1198– 1207.Morgan, J. L., Meier, R. P., & Newport, E. L. (1987). Structural packaging in the input to lan- guage learning: Contributions of prosodic and morphological marking of phrases to the acquisition of language. Cognitive Psychology, 19, 498–550.Morgan, J. L., Meier, R. P., & Newport, E. L. (1989). Facilitating the acquisition of syntax with cross-sentential cues to phrase structure. Journal of Memory and Language, 28, 360–374.Morgan, J. L., & Newport, E. L. (1981). The role of constituent structure in the induction of an artificial language. Journal of Verbal Learning and Verbal Behavior, 20, 67–85.Morgan, J. L., & Saffran, J. R. (1995). Emerging integration of sequential and suprasegmental information in preverbal speech segmentation. Child Development, 66, 911–936.Nazzi, T., Paterson, S., & Karmiloff-Smith, A. (2003) Early word segmentation by infants and toddlers with Williams syndrome. Infancy, 4, 251–271.Newport, E. L. (1990). Maturational constraints on language learning. Cognitive Science, 14, 11–28.Newport, E. L., & Aslin, R. N. (2004). Learning at a distance: I. Statistical learning of non- adjacent dependencies. Cognitive Psychology, 48, 127–162.Newport, E. L., Hauser, M. D., Spaepen, G., & Aslin, R. N. (2004). Learning at a distance: II. Statistical learning of non-adjacent dependencies in a non-human primate. Cognitive Psychology, 49, 85–117.Oakes, L. M., & Spalding, T. L. (1997). The role of exemplar distribution in infants’ differentia- tion of categories. Infant Behavior and Development, 20, 457–475.Papçun, G., Krashen, D., Terbeek, D., Remington, R., & Harshman, R. (1974). Is the left hemisphere specialized for speech, language, and/or something else? Journal of the Acoustical Society of America, 55, 319–327.Peña, M., Maki, A., Kovacic, D., Dehaene-Lambertz, G., Koizumi, H., Bouquet, F., & Mehler, J. (2003). Sounds and silence: an optical topography study of language recognition at birth. Proceedings of the National Academy of Sciences of the USA, 100, 11702–11705.Peretz, I. (in press). Introduction to The Nature of Music: A Special Issue of Cognition.Pinker, S. (1984). Language learnability and language development. Cambridge, MA: Harvard University Press.Pinker, S. (1985). Language learnability and children’s language: A multifaceted approach. In K. E. Nelson (Ed.), Children’s language: Vol. 5 (pp. 399–442). Hillsdale, NJ: Erlbaum.Pinker, S. (1991). Rules of language. Science, 253, 530–535.Pinker, S. (1999). Words and rules. London: Weidenfeld & Nicolson.Pisoni, D. B. (1975). Auditory short-term memory and vowel perception. Memory and Cognition, 3, 7–18.Plunkett, K., & Schafer, G. (1999). Early speech perception and word learning. In M. Barrett (Ed.), The development of language (pp. 51– 71). Hove: Psychology Press.Pollak, S. D., & Kistler, D. J. (2002). Early experience is associated with the development of categorical representations for facial expressions of emotion. Proceedings of the National Academy of Sciences of the USA, 99, 9072–9076.Quinn, P. C., & Eimas, P. D. (2000). The emergence of category development during infancy: Are separate and conceptual processes required? Journal of Cognition and Development, 1,[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... 55–61. 95. Domain-General Learning Capacities 85Redington, M., Chater, N., & Finch, S. (1998). Distributional information: A powerful cue for acquiring syntactic categories. Cognitive Science, 22, 425–469.Rice, M. L. (1999). Specific grammatical limitations in children with specific language impair- ment. In H. Tager-Flusberg (Ed.), Neurodevelopmental disorders (pp. 331– 359). Cambridge, MA: MIT Press.Rossen, M. L., Jones, W., Wang, P. P., & Klima, E. S. (1995). Face processing: Remarkable sparing in Williams syndrome. Genetic Counseling, 6, 138–140.Roy, D., & Pentland, A. (2002). Learning words from sights and sounds: A computational model. Cognitive Science, 26, 113–146.Sabbagh, M. A., & Gelman, S. A. (2000). Buzzsaws and blueprints: What children need (or don’t need) to learn language. Journal of Child Language, 27, 715–726.Saffran, J. R. (2001a). Words in a sea of sounds: The output of statistical learning. Cognition, 81, 149–169.Saffran, J. R. (2001b). The use of predictive dependencies in language learning. Journal of Memory and Language, 44, 493–515.Saffran, J. R. (2002). Constraints on statistical language learning. Journal of Memory and Language, 47, 172– 196.Saffran, J. R. (2003a). Statistical language learning: Mechanisms and constraints. Current Directions in Psychological Science, 12, 110–114.Saffran, J. R. (2003b). Absolute pitch in infancy and adulthood: The role of tonal structure. Developmental Science, 6, 37–45.Saffran, J. R., Aslin, R. N., & Newport, E. L. (1996). Statistical learning by 8-month-old infants. Science, 274, 1926–1928.Saffran, J. R., & Griepentrog, G. J. (2001). Absolute pitch in infant auditory learning: Evidence for developmental reorganization. Developmental Psychology, 37, 74–85.Saffran, J. R., Johnson, E. K., Aslin, R. N., & Newport, E. L. (1999). Statistical learning of tone sequences by human infants and adults. Cognition, 70, 27–52.Saffran, J. R., Pollak, S. D., Seibel, R. L., & Shkolnik, A. (in press). Dog is a dog is a dog: Infant rule learning is not specific to language. Cognition.Saffran, J. R., Reeck, K., Niehbur, A., & Wilson, D. P. (2005). Changing the tune: Absolute and relative pitch processing by adults and infants. Developmental Science, 8, 1–7.Saldaña, H. M., & Rosenblum, L. D. (1993). Visual influences on auditory pluck and bow judg- ments. Perception and Psychophysics, 54, 406–416.Samuelson, L. K. (2002). Statistical regularities in vocabulary guide language acquisition in connectionist models and 15–20-month-olds. Developmental Psychology, 38, 1016–1037.Schouten, M. E., & van Hessen, A. J. (1992). Modeling phoneme perception: I. Categorical perception. Journal of the Acoustical Society of America, 92, 1841–1855.Shatz, M. (1994). Review of Laura and First Verbs. Language, 70, 789–796.Senghas, A., Kita, S., & Ozyurek, A. (2004). Children creating core properties of language: Evidence from an emerging sign language in Nicaragua. Science, 305, 1779–1782.Skinner, B. F. (1957). Verbal Behavior. New York: Appleton-Century-Crofts.Slater, A. (1995). Visual perception and memory at birth. In C. Rovee-Collier & L. P. Lipsitt (Eds.), Advances in infancy research: Vol. 9 (pp. 107–162). Norwood, NJ: Ables.Soja, N. N., Carey, S., & Spelke, E. S. (1991). Ontological categories[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... guide young children’s inductions of word meaning: Object terms and substance terms. Cognition, 38, 179–211.Swingley, D. (2005). Statistical clustering and the contents of the infant vocabulary. Cognitive Psychology, 50, 86–132.Thiessen, E. D., & Saffran, J. R. (2003). When cues collide: Use of stress and statistical cues to word boundaries by 7- to 9-month-old infants. Developmental Psychology, 39, 706–716. 96. 86 Jenny R. Saffran and Erik D. ThiessenTincoff, R., & Jusczyk, P. W. (1999). Some beginnings of word comprehension in 6-month-olds. Psychological Science, 10, 172–175.Vouloumanos, A., & Werker, J. F. (2004). Tuned to the signal: The privileged status of speech for young infants. Developmental Science, 7, 270–276.Wang, X., & Kadia, S. C. (2001). Differential representation of species-specific primate vocaliza- tions in the auditory cortices of marmoset and cat. Journal of Neurophysiology, 86, 2616– 2620.Wang, X., Merzenich, M. M., Beitel, R., & Schreiner, C. E. (1995). Representations of species- specific vocalization in the primary auditory cortex of the common marmoset: Temporal and spectral characteristics. Journal of Neurophysiology, 74, 2685–2706.Waxman, S. R., & Booth, A. E. (2000). Principles that are invoked in the acquisition of words, but not facts. Cognition, 77, B33–B43.Weismer, S. E., Evans, J., & Hesketh, L. J. (1999). An examination of verbal working memory in children with specific language impairment. Journal of Speech, Language, and Hearing Research, 42, 1249– 1260.Werker, J. F., Cohen, L. B., Lloyd, V. L., Casasola, M., & Stager, C. L. (1998). Acquisition of word–object associations by 14-month-old infants. Developmental Psychology, 34, 1289–1309.Younger, B. A., & Cohen, L. B. (1986). Developmental changes in infants’ perception of correla- tions among attributes. Child Development, 57, 803–815.Zatorre, R. J. (2001). The biological foundations of music (Eds. R. J. Zatorre & I. Peretz). Annals of the New York Academy of Sciences, 930.Zatorre, R. J., Belin, P., & Penhune, V. B. (2002). Structure and function of auditory cortex: music and speech. Trends in Cognitive Sciences, 6, 37–46. 97. 5How Inherently Social is Language?Dare Baldwin and Meredith MeyerParadoxically, the claim that social capacities are fundamental to language acquisitionsimply states the obvious and yet at one and the same time sparks considerable contro-versy. On the banal side, virtually no one would deny that language is inherently asocial phenomenon, and thus that the acquisition of language rides in some form onsocial factors. Controversy arises, however, when social factors are broadly accorded apivotal role in shaping language acquisition. That is, highly prominent accounts ofcertain aspects of language acquisition – most notably the acquisition of structuralaspects of the language (e.g., grammar, phonology) – have traditionally concentrated onnon-social mechanisms as core driving forces behind acquisition. Some have begun toquestion this view in recent years, however, proposing that social mechanisms are centralto acquisition of structural aspects of language in addition to other components, suchas the[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... lexicon. One way to conceptualize the history of research on language acquisition over thepast four or five decades is in terms of a halting but inexorable thrust toward making acrucial role for social capacities increasingly explicit and rigorous. In this chapter we willpursue three specific goals: to trace the checkered history of researchers’ attention tosocial factors in accounts of language acquisition, to elucidate some of the specific wayssocial capacities are now known to facilitate language acquisition, and to articulate howa socially based account of language acquisition has recently gained momentum to giverise to the high level of controversy that currently surrounds the topic.Starting DistinctionsGaining clarity on debates concerning the role of social factors in language acquisitionrequires distinguishing three kinds of social factors: social input, social responsiveness, 98. 88 Dare Baldwin and Meredith Meyerand social understanding. Regarding social input, all agree that language learning takesplace only in a social milieu; that is, language learning obviously depends on the pres-ence of linguistic input, which, by definition, is social. However, precisely what formsocial and linguistic input must take for various aspects of language to emerge is not yetclear, and a range of interesting questions arises in this regard (see, e.g., Goldin-Meadow,2003, for relevant discussion). Another issue concerns infants’ and children’s responsiveness to social input.Language learning is a very different enterprise depending on what form such respon-siveness takes. An example here is infants’ responsiveness to “motherese” (or, moreappropriately, infant-directed talk). This responsiveness benefits their acquisition, inthat infant-directed talk provides exaggerated clues to meaning and structure nowknown to aid learning (e.g., Fernald, 1989; Jusczyk, 1997; Kuhl, 2004). Regardingthe general topic of social responsiveness, it will be important to determine specificallywhat kinds of social responsiveness are key to language learning, as well as which spe-cific aspects of language learning depend heavily on fundamental forms of socialresponsiveness. Finally, social understanding (also known as social cognition) – skill at interpretingpeople’s desires, intentions, and beliefs – also plays a role in language learning (e.g.,Baldwin, 2000; Tomasello, 1999). Here the questions concern precisely what kind ofsocial understanding language learners – characteristically infants and young children– must possess for language acquisition to progress normally, how advances in socialunderstanding might in turn enhance (and, interestingly, in some cases perhaps compli-cate) language learning, and which aspects of language acquisition (e.g., word meaningsvs. grammatical structure) hinge on early- emerging social understanding. In what followswe will consider the role played by each of these three kinds of social factors – propertiesof the input, responsiveness to the input, and use of “people smarts” to mine the input– in children’s success at two of the basic tasks of language acquisition: discovering wordmeanings, and acquiring grammar.Discovering Word MeaningsThe meaning and reference components of language are social through and through, inthe sense that words themselves don’t actually mean[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... or refer to anything; they mean orrefer only by virtue of being used by people to mean or refer (e.g., Lyons, 1977; Quine,1960). If we hear new words emitted by loudspeakers at random intervals we can’t intuittheir meanings; conversely, if we encounter new things in the world, no degree of carefulinspection of the objects themselves will enable us to divine relevant labels for thesethings. We need to be around people using words meaningfully to discern referenceand acquire word meanings. The discovery of meaning is thus inherently an extendedprocess of social coordination. What form does this social coordination process take?All three dimensions delineated earlier – social input, social responsiveness, and socialunderstanding – are directly relevant here. 99. How Inherently Social is Language? 89Social input regarding word meaningFrom birth infants are immersed in a social milieu rich with meaning- relevant informa-tion. The language they hear is rife with multimodal clues to meaning; this is especiallytrue of the language that speakers in many cultures direct toward infants – now calledinfant-directed talk (IDT). Infant-directed talk is a meaning-rich stimulus along manydimensions. Among other things, when speaking to infants adults are more likely to uttercontent words (e.g., object labels, verbs, adjectives) in isolation (Brent & Siskind, 2001),to place content words in sentence-final position and at pitch peaks (Snow & Ferguson,1977; Stern, Speiker, & MacKain, 1982), and to talk about things in the immediate,here-and-now context (Snow & Ferguson, 1977). These modifications have the potentialto assist infants in extracting the relevant portion of the continuous sound stream (e.g.,a specific word) and to link it with an object or event in the immediate surround. Adults also tend to exaggerate intonation in ways that are correlated with emotionalcontent and communicative intentions (e.g., Fernald, 1989). For example, mothersspeaking languages quite diverse in typology (e.g., American English, Japanese, Hausa)all display a similar set of distinctive intonational patterns when attempting to conveyspecific emotion-laden messages to their infants (e.g., low-pitched, fluid intonation forsoothing; low-pitched, staccato intonation for prohibiting; high-pitched, rapid-excursionintonation for attentional enhancement). By virtue of their use of these meaning-ladenintonational contours, mothers are potentially providing infants with access to meaninglong before infants have come to be able to interpret the conventional meanings of thespecific words involved. Infants of course have access to prosodic/intonational propertiesof language well before they are born; hence even fetuses have the opportunity to beginprocessing this meaning-relevant, socially rich dimension of speech. Adults – at least in Western, middle-class culture – also expend effort to achieveattentional coordination with infants in ways that ought to facilitate infants’ discoveryof meaning. For example, adults often follow in on infants’ attentional focus and providelanguage relevant to that focus (e.g., Collis, 1977; Harris, Jones, & Grant, 1983;Tomasello & Todd, 1983). This is far from easy because infants’ attention is mercurial,and not surprisingly such attempts at follow-in labeling meet with less than[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... perfectsuccess. As well, when not following infants’ attentional lead, Western, middle-classadults show high rates of actively trying to direct infants’ attention toward referentsunder discussion via gestures such as pointing and showing (e.g., Akhtar, Dunham, &Dunham, 1991; Kaye, 1982). Not surprisingly, adults’ efforts at social coordination asthey engage with infants clearly matter for meaning acquisition. Infants of mothers whoengage in higher rates of follow- in labeling progress faster in vocabulary acquisition (e.g.,Akhtar et al., 1991; Harris, Jones, Brookes, & Grant, 1986; Tomasello & Todd, 1983),and in experimental procedures infants often more readily learn to comprehend newobject labels when the labels are introduced in the context of follow-in labeling (e.g.,Baldwin, 1993; Dunham, Dunham, & Curwin, 1993; Tomasello & Farrar, 1986). Adults also structure infants’ lives and activities in ways that support the establish-ment of “scripts” or “formats”: predictable patterns of action and interaction that enableinfants to build expectations about which objects will be contacted and mentioned when 100. 90 Dare Baldwin and Meredith Meyerand for what purpose (e.g., Bruner, 1981; Nelson, 1985). Such scripted activity poten-tially helps to make transparent the meaning of any accompanying language. Put moregenerally, responsive parenting reveals itself in the language domain as well as in otherdomains of social interaction, and such responsiveness on parents’ part benefits wordlearning (e.g., Tamis-LeMonda, Bornstein, Baumwell, & Damast, 1996). On many levels, then, infants are on the receiving end of linguistic input that isriddled with social clues to meaning – one might be tempted to say that the wordsthemselves are the least of the meaning-relevant information available. Of course,however, all this social richness would be of little value for meaning acquisition if infantswere insensitive to it, or if the modifications adults make were somehow incompatiblewith the mechanisms infants deploy for acquiring meaning.Meaning-relevant social responsivenessWhile in utero, infants are already sensitive to a range of linguistic properties, includingat least some prosodic characteristics of their native language (e.g., Nazzi, Bertoncini, &Mehler, 1998) and they can encode prosodic details of specific passages of speech (e.g.,DeCaspar & Spence, 1986). Given this precocious sensitivity to prosodic aspects oflanguage, it is not surprising that infants adore IDT (e.g., Fernald, 1985; Fernald &Kuhl, 1987; Werker & McLeod, 1989), which, among other things, exaggerates prosodiccharacteristics of language. Evidence on many fronts now indicates that IDT benefitslanguage learning. For example, IDT facilitates infants’ analysis of the phonologicalproperties of their native language (e.g., Burnham, Kitamura, & Vollmer-Conna, 2002;Jusczyk, 1997; Kuhl et al., 1997; Liu, Kuhl, & Tsao, 2003), promoting their ability toencode and recognize words within ambient speech. As well, IDT assists infants inexploiting statistical regularities within continuous speech to identify word segments(Thiessen, Hill, & Saffran, 2005). Other IDT modifications – such as placing relevantcontent words at the ends of sentences or at pitch peaks – also dovetail nicely with infants’processing strategies. For example, children learn words more readily[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... when they occursentence final relative to sentence internal (Echols & Newport, 1992). Interestingly,responsiveness to IDT is not universal. Where it is lacking, however, disruptions in lan-guage development also tend to be observed. Autism – associated strongly with disrup-tions in language development – is a case in point here. As a group, individuals withautism prefer to listen to non-speech analogs of IDT rather than IDT itself, while nor- mally developing children reliably prefer IDT (Kuhl, Coffey-Corina, Padden, & Dawson,2005). Children with autism even prefer to hear many superimposed voices over theirown mother’s IDT speech, in contrast to normally developing children and children withnon-autistic developmental delay (Klin, 1991, 1992). Infants’ learning about phonetic properties of speech plays a crucial role in wordlearning, and recent evidence clarifies that phonetic learning is also modulated by socialresponsiveness. For example, Kuhl, Tsao, and Liu (2003) found that 12 sessions ofexposure to Mandarin enabled English-learning American 9-month-olds to maintaintheir sensitivity to a Mandarin phonetic contrast not found in English. However, thisoccurred only if infants directly interacted with the Mandarin speaker. Infants who 101. How Inherently Social is Language? 91watched a DVD of the Mandarin speaker (who was filmed while interacting with a dif-ferent baby) did not maintain sensitivity to the Mandarin phonetic contrast. Contingentsocial interaction also increases the frequency and maturity of infants’ own vocalizations(e.g., Goldstein, King, & West, 2003). Finally, Kuhl et al. (2005) report that a subsetof children with autism who showed a preference for IDT over non-speech analogs alsodisplayed brain electrical activity in response to a phonetic discrimination task that wassignificantly closer to the normal pattern than did autistic children who preferred non-speech analogs over IDT. Based on these and related findings, Kuhl (2004) suggests thatsocial factors serve a “gating” function for neural computation of linguistic stimuli. Fernald and others (e.g., Fernald, 1989; Bruner, 1983) have demonstrated that IDTis rich in emotional and intentional content as well as in phonetic information, andinfants seem to be sensitive to the emotional correlates of distinct intonational contourswithin IDT. In one study (Fernald, 1993), for instance, infants showed higher rates ofattention and smiling when hearing utterances couched in intonation characteristic ofapproval than when hearing utterances conveyed in intonation typical of prohibition.Strikingly, this effect generally held up even when infants were hearing IDT of motherswho spoke a different language than the one with which infants themselves were familiar(e.g., English-learning infants smiled and attended more to Italian mothers’ approvalstatements than prohibitions). Infants’ attunement to intonation potentially assists wordlearning in a variety of ways. In a very basic sense, infants seem to be pulling “meaning”– even if not conventional semantic meaning – out of the intonation itself; in Fernald’sterms, “the melody carries the message” (1989, p. 1505). Infants’ sensitivity to the specificemotional content of a particular utterance via its intonational properties can facilitatetheir interpretation of the utterance. For example, an infant who[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... responds to attention-bid intonation accompanying an utterance like “It’s a buzz-a-bee!” with increased atten-tion to the relevant object is at an advantage for successfully mapping the new word(buzz-a-bee) to its appropriate referent. Along these lines, other research documents thatsuch labeling utterances indeed enhance infants’ attention to objects, and to object cat-egories more specifically, adding to the plausibility of this story (Baldwin & Markman,1989; Waxman, 2003). Infants are also highly responsive to adults’ attempts to coordinate attentional focus,which greatly expedites language learning. Infants start to reliably follow others’ gazeand pointing gestures during the second half of the first year (e.g., Butterworth, 1991;Carpenter, Nagell, & Tomasello, 1998), and they appreciate the object-directedness ofothers’ gaze and pointing gestures as early as 12 months of age (e.g., Woodward, 2003;Woodward & Guajardo, 2002). Interestingly, infants at 12 months will follow “gaze”or imitate a sequence of motions only if the entity involved appears to be animate, whichthey diagnose via a combination of surface characteristics such as eyes and fur and theentity’s propensity to engage in contingent behavior (e.g., Johnson, 2000; Shimizu &Johnson, 2004). This is one indication that gaze following and attention to gestures asearly as 12 months represents a genuine form of social responsiveness, rather than anon-social instance of reflexive orienting. In any case, responsiveness to others’ gaze,gestures, and actions helps to ensure that infants are focused on relevant referents whenobjects and events are under discussion, helping them to link words with the correctthings in the world. 102. 92 Dare Baldwin and Meredith Meyer Children with autism typically display significant deficits and/or delays in followingothers’ gaze and attention- directing gestures, and in interpreting others’ emotional dis-plays and action (e.g., Baron-Cohen, Baldwin, & Crowson, 1997; Mundy, Sigman,Ungerer, & Sherman, 1986; Osterling, Dawson, & Munson, 2002). It is likely that thesedeficits in social responsiveness contribute significantly to the word- learning delay thatis typical of this developmental disorder.Discovering what people mean when they use wordsIf we think of meaning acquisition as an ongoing task of social coordination, one obviousquestion is whether, and to what degree, learners themselves actively pursue such socialcoordination in the service of meaning acquisition. In other words, is it correct to saythat young children, or even infants, are trying to figure out what people use words tomean? Are they actively seeking to discover others’ intentions and attentional focus inthe service of drawing inferences about linguistic reference and meaning? And if so, howskillful are they at it? In an influential review, Shatz (1983) considered these (amongother) questions at some length, concluding at the time that the available evidence pro-vided no resolution. Inspired by Shatz’ seminal analysis, several researchers embarked onresearch specifically to address such questions. At the inception of such research, a relatively new body of work on children’s develop-ing theories of mind was revealing what appeared to be genuine conceptual deficits ininfants’ and young preschoolers’ understanding of others’ mental life. This work sug-[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... gested that children younger than about 4 did not yet conceive of mentalistic notionssuch as beliefs, intentions, or attention (e.g., Moses & Flavell, 1990; Wellman, 1990;Wimmer & Perner, 1983). Such findings gave rise to general skepticism toward the ideathat young children, or even infants, might be skilled contributors to the kind of socialcoordination that would promote language learning. If children couldn’t yet understandnotions such as intention, attention, or belief, for example, how could they possibly beskilled at tracking others’ intentions and attentional focus to guide inferences about wordmeaning? On the other hand, language learning might provide an especially sensitivewindow on children’s emerging mentalistic understanding, as it is a genuine, real-worldtask in which children are fully immersed on an ongoing basis. Perhaps, then, children or even infants might display budding skills for mentalisticreasoning in the context of language learning tasks that would be more difficult to observein other kinds of traditional experimental tasks. But what would constitute evidence thatchildren make use of genuine social understanding in the service of language learning?The strategy researchers adopted was to identify language learning scenarios in whichthe relevant piece of mentalistic understanding would make all the difference for learning;in particular, scenarios in which a failure to understand (and make active use of theunderstanding) would put children at risk for making errors. In the first study of thiskind, Baldwin (1991) presented infants in two age groups (16–17, and 18–19 months)with new labels for a novel toy in the context of “discrepant labeling” – a commonlyoccurring everyday phenomenon (e.g., Collis, 1977) in which a word is heard preciselywhen the learner is focused on a different object than the one to which the speaker intends 103. How Inherently Social is Language? 93to refer. In this context, infants might fall prey to a word-learning error – linking thenew word with an inappropriate referent – because their attention was focused on thewrong object at the time they heard the new label. If, however, infants possess some skillsfor tracking others’ referential intentions or attentional focus, they might be able to avoidan error. That is, errors needn’t occur if infants understand that the speaker intends tolabel a specific object, and actively monitor the speaker for clues (e.g., gaze direction,body posture, voice direction) to the intended referent. In this study, infants in both agegroups displayed high rates of checking the speaker’s face in response to hearing novellabels, as though checking for clues to the speaker’s referent. Their performance on asubsequent comprehension test confirmed that they made use of such clues: When askedto “Find the modi!” infants across the 16- to 19-month span rarely showed systematicselection of the toy they themselves had been focused on when they had heard the label.They did not make the word-learning error that the discrepant labeling scenario put themat risk for. This skill at avoiding errors was reliable in the younger group, but detectablymore sophisticated in the older group. The younger group seemed to simply block amapping between the novel label and the novel toy during discrepant labeling (they per-formed at chance levels to comprehension[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... questions), while the older group went onebetter: they not only avoided the error, they actually succeeded in figuring out the correctreferent of the novel label (they selected the toy that the speaker had been focused on inthe discrepant labeling event). One might wonder whether the younger group’s chancelevel performance was just an inability to establish new word–object links more generally,but an important control comparison revealed above- chance comprehension performancewhen the speaker labeled the toy infants were focused on. Thus their response to discrep-ant labeling seemed genuinely to be a case of blocking a potential error, whereas the 18- to19- month-olds were able to exploit the available clues more fully not only to avoid anerror but also to discover the correct referent of the new word. All in all, the findings of this initial study confirmed that infants actively and spon- taneously track clues to others’ referential intentions, and use these clues to draw infer-ences about word meanings. This was among the first pieces of evidence suggesting thepresence of some form of mentalistic understanding in infancy. These findings meshnicely with Shatz’ (1994) view that social understanding emerges early, but neverthelessundergoes gradual and extended development. The basic findings have since been repli-cated by others (e.g., Dunham et al., 1993; Hollich, Hirsh-Pasek, & Golinkoff, 2000),and elaborated in a number of ways (see Baldwin & Moses, 2001; Sabbagh & Baldwin,2005; Tomasello, 1999 for reviews), indicating that infants (a) actively exploit an impres-sive array of intentional clues to guide inferences about word meaning, (b) can trackthese clues across time in the context of relatively novel and complex interactive scenarios,(c) frequently take social clues indicating referential intent as criterial for establishingnew word–referent links (e.g., Baldwin et al., 1996), and (d) can mine input for socialclues to meaning even when overhearing language addressed to others (e.g., Akhtar,2005). Put another way, infants seem to track social clues to inform them about whichword-to-world correspondences are worthy of registering and recalling; in this sense,social clues serve a “gating” function for establishing new word meanings, which isreminiscent of the role Kuhl suggests social responsiveness plays in phonologicaldevelopment. 104. 94 Dare Baldwin and Meredith Meyer Jaswal (2004) has also demonstrated that young children’s intention detection skillsnot only guide their initial inferences about reference and meaning, but also facilitatetheir subsequent elaboration of word meaning. Finally, several recent studies have docu-mented that language-related intention-monitoring processes are disrupted in childrenwith autism. In particular, autistic children are prone to making word-learning errorsspecifically as a result of deficits in exploiting the social/intentional clues that speakersdisplay (Baron-Cohen et al., 1997; Preissler & Carey, 2005). This body of work on language-relevant social understanding has sparked controversyon a number of fronts. Two issues in particular have been the focus of considerabledebate. The first of these was whether social understanding is really operating after all(e.g., Hoff & Naigles, 2002; Ruffman, 2003; Samuelson & Smith, 1998). At this junc-ture, the[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... evidence is clear that infants nearing the end of their second year indeed spon-taneously and actively monitor clues indicative of others’ intentions, and use these cluesto guide inferences about word meaning (e.g., Diesendruck, Markson, Akhtar, & Reudor,2004; Hollich et al., 2000; Moore, Angelopoulos, & Bennett, 1999). On the other hand,there remains a real question whether this skill at capitalizing on social clues warrantsthe label “social understanding,” or whether it is simply an impressive demonstration ofinfants’ ability to exploit statistical regularities in others’ behavior to guide learning(e.g., Ruffman, 2003; see also Perner & Ruffman, 2005, and Saffran & Thiessen, thisvolume, for related discussion). Disentangling the operation of genuine inferences aboutgoals and intentions from sophisticated statistical learning is a challenging enterprise(Povinelli & Vonk, 2003) and remains a focus of current research. At present, however,it can safely be said that infants as young as 12 to 18 months are considerably moreactive in social coordination and more savvy in mining clues from others’ behavior toguide language learning than anyone had previously imagined. A second controversy hinged on the possibility that social understanding might actu-ally be responsible for word-learning skills previously thought to arise from other mecha-nisms, such as word-learning constraints (e.g., Bloom, 2000; Saylor, Baldwin, & Sabbagh,2004; Tomasello, 2000a; see Poulin-Dubois & Graham, this volume, regarding theconstraints approach). The basic idea here is that word-learning constraints such as thewhole object and mutual exclusivity assumptions (e.g., Markman, 1989) can readily becouched in terms of socially based assumptions on children’s part. For example, theassumption that new words refer to whole objects rather than to parts, properties, orevents (the whole object assumption) can readily be construed as the assumption thatwhen speakers use words they intend to refer to whole objects rather than parts, proper-ties, or events. Some recent data (a) indicate that parental input exhibits pragmatic regu-larities that would support children’s adoption over time of constraints such as themutual exclusivity assumption (Callanan & Sabbagh, 2004), and (b) point to pragmaticunderstanding providing a better account of children’s inferences about meaning thanconstraints construed in non-pragmatic terms (e.g., Diesendruck & Markson, 2001;Saylor, Sabbagh, & Baldwin, 2002, but see Preissler & Carey, 2005, for potentiallycontradictory evidence from children with autism). Summing up to this point, linguistic input is chock-full of clues to meaning; manyof these clues are fundamentally social in that they reside in intonation, facial expression,gaze, gesture, ongoing action, and the history of social interaction. Children’s social 105. How Inherently Social is Language? 95responsiveness and their growing ability to actively exploit such social sources of infor-mation contribute immensely to the smooth and speedy trajectory so characteristic ofearly word learning.Acquiring GrammarGiven that word meanings are so obviously a product of social consensus, a central rolefor social processes in meaning acquisition seems entirely sensible. What of anotherdomain equally important in acquiring language, namely the acquisition of syntax? Herethere is[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... substantial disagreement on how central a role social understanding can play.Specifically, a distinction in the syntactic domain can be drawn between formalist theo-ries stressing pre-adapted acquisition mechanisms dedicated exclusively to grammar andsocio-pragmatic accounts centering on grammar learning as a fundamentally socialenterprise.Nativist accounts background social input to linguistic structureChildren’s ability to discover latent grammatical structure in the input they encounterhas long been a source of inquiry. Formalist theories of syntax acquisition garneredwidespread support soon after their introduction in the 1960s, offering an attractivealternative to the then-dominant, and clearly deficient, behaviorist version of “learned”language (e.g., Chomsky, 1959). Formalist proposals account for the acquisition ofstructural properties of language via universal constraints specific to syntax (see Lidz,this volume). On this view, syntactic competence is enabled by universal grammar (UG),an innate body of abstract, highly generalizable grammatical representations. This per-spective necessitates a strongly nativist and modular account of syntax; consequentlyresearchers allying themselves with a formalist position tend to question whether socialcapacities could play any sort of substantial role in syntax acquisition (e.g., Fisher, 2002;Lidz & Waxman, 2004). Such an outlook acknowledges social factors insofar as it rec-ognizes that language acquisition depends crucially on the learner having access to alinguistic (thus, by definition, social) environment rich in positive evidence, but oncesuch an environment is provided to the learner, the acquisition process is seen to relymost heavily on the operation of innate structural principles that are triggered by input.In other words, the formalist view gives little more than a reflexive nod in the directionof the social milieu within which language is acquired. The formalist conception of a functionally independent, modular system of syntaxleaves little room for investigation of social capacities in the domain of syntax acquisition.Indeed, a major reason that UG is considered the only plausible solution for syntacticcompetence is a belief that the input that children receive from the social world is incom-plete; in particular, linguistic input is thought to lack clear evidence about the syntacticacceptability of utterances (e.g., Hyams, 1986; Pinker, 1994). Adults are less likely toprovide negative feedback for syntactic errors than for violations of truth (Brown & 106. 96 Dare Baldwin and Meredith MeyerHanlon, 1970); furthermore, the feedback that is provided for ungrammatical utterancesis rarely consistent and is regarded by many to be insufficient as a source of reliableinformation about syntactic well-formedness (e.g., Marcus, 1993). Despite such anapparently impoverished stimulus, young children just beginning to combine wordsroutinely avoid certain errors, and errors they do display seem principled, indicating theoperation of an underlying system. For example, children rarely make errors in the useof auxiliaries that would be expected if they were simply generalizing from adult sentencepatterns with main verbs (Stromswold, 1990, as cited in Pinker, 1994), and even 1-year-old children seem to operate with adult-like rules governing how adjectives may[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... modifycertain types of noun phrases (Bloom, 1990). Such patterns provide support for theformalist claim that even very early speech reflects underlying syntactic competence.Initial socio-pragmatic alternatives emphasize social support forlanguage acquisitionSoon after Chomsky articulated the first version of a generative grammar, a number ofdevelopmentalists sympathetic to pragmatic views of human behavior raised objectionsto what they viewed as a neglect of social factors on the part of formalist theories.According to the formalist position, a learner is already in possession of structural rep- resentations by virtue of UG, independent of any social experience in the world. Prag-matic developmentalists criticized what they believed to be an overly narrow focus onthe innate aspects of structure, arguing that theories of language needed to incorporatesocial features and the ways in which children could respond to and learn from adults(Bruner, 1975; Halliday, 1975). Bruner offered a particularly influential argument in this vein, proposing that gram- matical knowledge emerged in the context of a child’s linguistic interactions with an adult,particularly during communicative exchanges taking place during play (Bruner, 1975,1981). He described a language assistance system (LAS), which encompassed a develop-mentally sequenced shift in communicative “initiative.” Such initiative rested first withadults and was then progressively transferred to children via structured “formats” providedduring play or other one-on-one interaction; children would gradually come to internalizethe meaning and effect of a particular speech act, including the structural aspects of suchan act. For example, Bruner proposed that children acquired basic order rules, at leastin part, by observing repeated adult-narrated enactments of an agent–action–object–recipient relationship. In Bruner’s conceptualization of syntactic development, then, basicsocial activities that jointly engage child and adult are crucial in providing the contextualsupport within which structure is acquired. Although LAS was not advanced as a mutu-ally exclusive alternative to the Chomskyan language acquisition device (LAD), othertheorists would later be influenced by Bruner’s suggestion that LAS might play a largerole in linking structure with the conceptual relationships expressed by grammar. Bruner’s ideas did much to encourage a deeper consideration of social factors whenconsidering the task of language acquisition in general. An examination specific to gram-matical development, however, rendered Bruner’s account of syntax problematic forseveral reasons. First, interactions in the social sphere that Bruner proposed to be a source 107. How Inherently Social is Language? 97of grammatical understanding (such as the agent–action–object–recipient relationship)did not appear to provide a transparent way to directly map conceptual relationshipsonto meaningful word order (Shatz, 1981; Slobin, 1982). Second, contrary to whatBruner would predict, the course of syntactic development appeared largely uninfluencedby differences in available social support. For example, variability in maternal speech-style along dimensions such as mean length of utterance and propositional complexitywas found to be largely unrelated to child language development (Newport, Gleitman,& Gleitman, 1977).[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Furthermore, studies of blind children lacking access to much ofthe social information available in such behavior as joint eye gaze and pointing neverthe-less evidenced a nearly normal progression of syntactic development (Gleitman &Newport, 1995; Landau & Gleitman, 1985). Similarly, examinations of the developmentof gestures in deaf children deprived of normal linguistic input from their hearing parentssuggested a remarkable resiliency in the emergence of structured communication (e.g.,Goldin- Meadow, 1985, 2003).Contemporary socio-pragmatic accounts: Emphasizing children’ssocial understandingBruner’s writings endorsed the idea that linguistic support provided during one-on-onesocial contact with a parent or other adult is crucial to acquiring all aspects of language.The fact that substantial variability in the social context of language learning seemed tomatter little in syntactic development thus posed difficulties for Bruner’s account of howchildren come to understand and produce grammatical speech. More recent socio-pragmatic accounts of syntax acquisition address this problem by appealing more heavilyto the ways in which children capitalize on their own social abilities in the service ofacquisition – in other words, they focus more on the social knowledge that childrenbring to bear in the acquisition of language. By crediting the learner with an ability tocapitalize on social knowledge from the very start, rather than positing the adult as thesole source of social support, contemporary accounts envision the child as an active andskilled participant in the language learning task. A child operating on the assumptionthat others intentionally use language to communicate has a substantial leg up on theclassic Brunerian child, who is benefiting in a more passive sense from parental socialsupport. This crucial difference lends current accounts considerably more strength andat the same time opens the door to an even more central role for social capacities in theacquisition of structure; social capacities are seen as a property of the child, and theinformation available to solve the acquisition task is considered to be the outcome ofboth the socially skilled child and the socially rich environment. In rejecting the claim that children acquire syntax using innate, pre-specified gram-matical representations, socio-pragmatic theorists argue that children learn grammaticalstructure by observing and analyzing adults’ usage (e.g., Tomasello, 2003, 2004). Theacquisition of both structure and content is seen to rest fundamentally on children’sability to draw inferences about the adult’s communicative intent to acquire meaning –crucially, meaning conveyed not just by words, but by grammatical structure itself.The process is not strictly imitative, however; in addition to capitalizing on their 108. 98 Dare Baldwin and Meredith Meyerunderstanding of the function of any one particular speech act, children are believed toadditionally recruit domain-general skills such as statistical learning and pattern recogni-tion (e.g., Saffran, Aslin, & Newport, 1996) as well as structure mapping (e.g., Gentner,Holyoak, & Kokinov, 2001). Generativity, the feature of language that first promptedformalists to posit universal grammar, is thus explained not by a domain-specific lan-guage module, but rather by domain-general processes[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... that operate in conjunction withan understanding of referential and communicative intent. Current socio-pragmatic accounts thus incorporate social knowledge on the part ofthe young learners themselves, a shift from the early Brunerian focus on the externalsupport that a parent offers. These contemporary accounts also differ from earlier socio-pragmatic work in that they explicitly describe syntax itself in very different terms fromthose of formalist theories. As a result, current socio-pragmatic accounts of syntax aremore finely articulated than the original versions advanced by pragmatic developmental-ists in the 1970s. They gain clarity by drawing heavily from theories of cognitive andusage-based grammars arising from a group of linguists skeptical of formalist claims ofthe logical necessity of UG (e.g., Bybee, 1985; Givón, 1993; Langacker, 1987; Van Valin,1993). According to these theories, syntax reflects the tendency to regularize repeatedpatterns of word usage into grammatical structure. It is the tendency to regularize (gram- maticize) that is universal across human cultures, not the syntactic representationsthemselves. As a consequence, socio-pragmatic accounts of language characterize com-petence as being in possession of an “inventory” of language-specific constructions ratherthan in possession of abstract knowledge of syntax. Both formalist and socio-pragmatic accounts include explanations for many patternsobserved in naturalistic speech data. As discussed above, the finding that children “playby the rules” of the syntax of their native language has traditionally been used to arguethat humans come equipped with abstract grammatical representations to the languagelearning task. Socio-pragmatic accounts do not deny that children’s speech often obeysgrammatical convention (although see Tomasello, 2000b, for important counterexam-ples) or that children may show evidence of understanding structure before they canproduce it, but they reject the claim that these findings are evidence of UG. If childrenindeed appreciate that language is used to communicate, then it is unsurprising thattheir own productions resemble adults’; they should naturally produce the forms thatothers use to fulfill specific communicative goals.Empirical evidence for social capacities in syntaxEmpirical investigations of syntax development shed additional light on the mechanismsinvolved in the acquisition of structure. Demonstrations of early syntactic understanding(e.g., Fisher, 2000; Hirsh-Pasek, Golinkoff, & Naigles, 1996) have traditionally beenprovided as strong support for theories of universal grammar. For example, children asyoung as 25 months are able to distinguish between transitive and intransitive sentenceframes and can use this difference to infer the meaning of a novel verb (Naigles, 1990).This finding suggests both that children possess crucial syntactic knowledge (i.e., thetransitive/intransitive distinction) at a remarkably early age and that they can use such 109. How Inherently Social is Language? 99syntactic knowledge to guide inferences about verb meaning. Socio-pragmatic theorists,however, explain these and similar data according to a model in which no universalgrammar is[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... necessary; rather, children are argued to possess such understanding becausethey have acquired the structure from adult examples through the same mechanismsused to acquire meaning, namely an understanding of communicative intent coupledwith domain-general learning mechanisms that can allow for generalization to novelexamples. In a recent synthesis of both naturalistic speech production and experimental data,Tomasello (2000b) outlines a “verb-island” hypothesis, which stands in stark oppositionto the formalist idea that children come equipped for syntax acquisition with abstractgrammatical knowledge. This hypothesis states that very young children initially acquireverbs piecemeal by producing forms as they are received from the input. Later, once abody of verbs has been acquired, children can engage in off-line analysis of structuraland semantic relationships across verbs, gradually constructing abstract syntactic usagerules over a period of months or years. Tomasello argues that an examination of youngchildren’s speech demonstrates a marked reluctance on children’s part to go beyondexamples provided in the input (e.g., Tomasello, 1992, 2000b). Such conservatism istaken as evidence counter to what a formalist account would predict: a theory of abstractsyntactic knowledge predicts children will apply syntactic knowledge “across the board”to verbs as a category rather than on a case-by-case basis. Finally, Tomasello further notesthat studies indicating comprehension of abstract grammatical structure, such as Naigles’(1990) study of transitive and intransitive verbs, tend to be conducted on populationsof children who have been exposed to years of input. Without directly assessing theinformativeness of this input, then, any ability evidenced by these children could plau-sibly be argued to have been acquired through social means. In response to Tomasello’s account that children acquire verbs piecemeal rather thanby making use of an underlying innate grammar, Fisher (2002) counters that findingssuch as children’s reluctance to innovate and uneven morphological marking are in noway problematic for formalist accounts. Although such theories do call for abstractknowledge that aids children in applying syntactic rules to novel instances, Fisher alsoreminds us that such theories fully recognize the importance of experiencing a language-specific pattern of verb usage. Tomasello’s socio-pragmatic approach has also been criti-cized for relying too heavily on a unidirectional relationship between social capacitiesand linguistic competence. Shatz (1992), for example, encourages viewing the two abili-ties as mutually facilitative rather than positing social ability as the precursor and maincontributor to language. Finally, socio-pragmatic theories are also sometimes believed tobe hard-pressed to explain a large body of evidence describing dissociations betweensocial abilities and grammatical competence (for a discussion of this and related issuessee Shatz, 1992, and Tomasello, 1995). For empirical evidence to be definitive in this debate, formalists and socio-pragmatictheorists will need to reach consensus on contrasting predictions that these accountsmake (see Akhtar, 2004, for discussion on this topic). Lidz, Waxman, and Freedman(2003) argue that they[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... have done exactly that. Their claims rest in part on a centraldifference between formalist and socio-pragmatic accounts, namely these theories’ posi- tions on the availability of evidence in the input. Children’s acquisition of structure on 110. 100 Dare Baldwin and Meredith Meyera socio-pragmatic account depends crucially on the relevant evidence for that structurebeing available in the input. By contrast, the pre-specified abstract syntactic knowledgewith which children are gifted on a formalist account could enable them to acquire somestructures for which the input fails to provide sufficient evidence. Lidz et al. (2003) found that young children possess abstract knowledge of hierarchi- cally structured noun phrases despite minimal input demonstrating such structure.Specifically, 18-month-old children were shown to have an understanding that anaphoricone refers to a constituent consisting of an adjective and a noun rather than the nounalone. Infants possessed this understanding despite the fact that relevant examples ofanaphoric one were rare in parental input (see Lidz, this volume). If Lidz and colleagues’analysis is correct in claiming that children possess syntactic knowledge despite a povertyof the stimulus, then evidence that children can acquire anaphoric one is tantamount toevidence of universal grammar in action. Debate has arisen regarding the viability of thepoverty-of-the-stimulus argument here, however. Akhtar, Callanan, Pullum, and Scholz(2004) not only dispute the claim that positive examples of anaphoric one are rare, butalso raise the possibility that pragmatic inference may aid a child in figuring out thecorrect referent of anaphoric one (see also Lidz & Waxman, 2004; Tomasello, 2004). Itis exactly this type of ability – skill at inferring the communicative goals of others – thatis crucial to a socio-pragmatic account of syntax acquisition. Clearly, the findings as they currently stand invite future inquiry into processesunderlying the acquisition of syntax. At present, evidence for innate structure guidingsyntax acquisition is in dispute. Yet at the same time, no actual direct evidence docu-ments the specific role of intention detection in children’s processing of structural aspectsof others’ language use. Pursuing such evidence is one fruitful approach to resolving thecurrent debate.Wrapping Up: Social Factors Influence Language Learning atMany Levels of AnalysisTo sum up our discussion, we now know that social processes play a multitude of rolesin language learning. Regarding children’s discovery of meaning, language input tochildren is infused with a rich array of social clues to meaning, and existing researchindicates that infants and young children are highly responsive to a variety of these clues.Put another way, children – who, when developing normally, are highly attuned to socialclues – are immersed from day one in a social context that shapes their experience inways that make relevant referents salient and appropriate meanings transparent. At thesame time, from as early as 12 months of age, children themselves actively monitormeaning-relevant social clues that others exhibit, and put these clues to work to guideinferences about reference and meaning. Powerful skills for statistical tracking availableat least as early as 7 to 8[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... months likely play an important role in how children accomplishthis, but at some point – quite possibly as early as 12 to 18 months – children spontane- ously track others’ social clues because they understand these clues offer them a usefulwindow on others’ referential intentions. Some of children’s pragmatic inferences – those 111. How Inherently Social is Language? 101that are strongly supported across a wide range of contexts – may undergo a process ofautomatization that crystallizes them into default assumptions (constraints) that theycan deploy to drive inferences about meaning even in the absence of available on-linesocial clues. All in all, children seem to approach the discovery of meaning as a socialpuzzle, and skillfully mine their social surroundings for clues to solve the puzzle. Regarding meaning acquisition, few have ever doubted a central role for socio-pragmatic factors, although the precise nature of this role has generated controversy fromtime to time. By contrast, socio- pragmatic factors have frequently been dismissed aslargely irrelevant to the acquisition of syntax. However, like the mythical phoenix, socio-pragmatic factors have recently re-emerged from the ashes of syntactic immolation. Earlysocio-pragmatic accounts emphasized the role that richly structured social input playsin grammar acquisition. Such accounts languished in the face of evidence showcasingthe robustness of syntax acquisition despite substantial variability in the social contextof language learning. Akhtar and Tomasello have recently breathed new life into a socio-pragmatic account for syntax acquisition. Their account emphasizes the new body ofevidence documenting early-emerging social understanding; they propose that childrenactively work to discern the communicative purpose of structural patterns in the lan-guage. Children’s skills for intention-processing, coupled with statistical learning, patternrecognition, and structure-mapping abilities, may be what enable them to discover latentstructural principles underlying language use. Once again, socio-pragmatic factors arestrong contenders for children’s discovery of grammatical structure. This newly enlivenedsocio- pragmatic account presents a challenge to formalist accounts that is already receiv-ing considerable attention. At the same time, the full power of this resurrected socio-pragmatic account is far from resolved. Whether it can account for the full range ofcomplex syntactic structures that children acquire remains an important avenue forempirical investigation. At the very least, however, recent evidence makes it now safe toventure that social factors influence language learning at many levels of analysis – struc-tural (phonological and syntactic) as well as content-related (word meaning). Languageis inherently social to a very deep degree.NoteWe thank the volume editors for supporting our work on this chapter. The manuscript wasprepared in part by means of funds provided by the National Science Foundation under GrantNo. BCS-0214484.ReferencesAkhtar, N. (2004). Nativist versus constructivist goals in studying child language. Journal of Child Language, 31, 459–462.Akhtar, N. (2005). The robustness of learning through overhearing. Developmental Science, 8, 199–209.Akhtar, N., Callanan, M., Pullum, B., &[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Scholz, G. (2004). Learning antecedents for anaphoric one. Cognition, 93, 141–145. 112. 102 Dare Baldwin and Meredith MeyerAkhtar, N., Dunham, F., & Dunham, P. J. (1991). Directive interactions and early vocabulary development: The role of joint attentional focus. Journal of Child Language, 18, 41–49.Baldwin, D. A. (1991). Infants’ contribution to the achievement of joint reference. Child Develop- ment, 62, 875–890.Baldwin, D. A. (1993). Infants’ ability to consult the speaker for clues to word reference. Journal of Child Language, 20, 395–418.Baldwin, D. A. (2000). Interpersonal understanding fuels knowledge acquisition. Current Directions in Psychological Science, 9, 40–45.Baldwin, D. A., & Markman, E. M. (1989). Establishing word–object relations: A first step. Child Development, 60, 381– 398.Baldwin, D. A., Markman, E. M., Bill, B., Desjardins, R. N., Irwin, J., & Tidball, G. (1996). Infants’ reliance on a social criterion for establishing word– object relations. Child Development, 67, 3135–3153.Baldwin, D. A., & Moses, L. J. (2001). Links between social understanding and early word learning: Challenges to current accounts. Social Development, 10, 309–329.Baron- Cohen, S., Baldwin, D. A., & Crowson, M. (1997). Do children with autism use the speaker’s direction of gaze strategy to crack the code of language? Child Development, 68, 48–57.Bloom, P. (1990). Syntactic distinctions in child language. Journal of Child Language, 17, 343–355.Bloom, P. (2000). How children learn the meanings of words. Cambridge, MA: MIT Press.Brent, M. R., & Siskind, J. M. (2001). The role of exposure to isolated words in early vocabulary development. Cognition, 81, B33–B44.Brown, R., & Hanlon, C. (1970). Derivational complexity and the order of acquisition in child speech. In J. R. Hayes (Ed.), Cognition and the development of language (pp. 11–54). New York: Wiley.Bruner, J. S. (1975). The ontogenesis of speech acts. Journal of Child Language, 2, 1–19.Bruner, J. S. (1981). The social context of language acquisition. Language and Communication, 1, 155–178.Bruner, J. S. (1983). Child’s talk: Learning to use language. New York: W. W. Norton.Burnham, D., Kitamura, C., & Vollmer-Conna, U. (2002). What’s new, pussycat? On talking to babies and animals. Science, 296, 1435.Butterworth, G. (1991). The ontogeny and phylogeny of joint visual attention. In A. Whiten (Ed.), Natural theories of mind (pp. 223–232). Oxford: Basil Blackwell.Bybee, J. (1985). Morphology. Amsterdam: Benjamins.Callanan, M. A., & Sabbagh, M. A. (2004). Multiple labels for objects in conversations with young children: Parents’ language and children’s developing expectations about word mean- ings. Developmental Psychology, 40, 746–763.Carpenter, M., Nagell, K., & Tomasello, M. (1998). Social cognition, joint attention, and com- municative competence from 9–15 months of age. Monographs of the Society for Research in Child Development, 63 (4, Serial No. 255).Chomsky, N. (1959). A review of B. F. Skinner’s “Verbal Behavior.” Language, 35, 26–58.Collis, G. M. (1977). Visual co-orientation and maternal speech. In H. R. Schaffer (Ed.), Studies in mother–infant interaction (pp. 355–375). New York: Academic Press.DeCaspar, A. J., & Spence, M. J. (1986). Prenatal maternal speech[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... influences newborns’ percep- tion of speech sounds. Infant Behavior and Development, 9, 133–150.Diesendruck, G., & Markson, L. (2001). Children’s avoidance of lexical overlap: A pragmatic account. Developmental Psychology, 37, 630–641. 113. How Inherently Social is Language? 103Diesendruck, G., Markson, L., Akhtar, N., & Reudor, A. (2004). Two-year-olds’ sensitivity to speakers’ intention: An alternative account of Samuelson and Smith. Developmental Science, 7, 33–41.Dunham, P. J., Dunham, F., & Curwin, A. (1993). Joint attentional states and lexical acquisition at 18 months. Developmental Psychology, 29, 827–831.Echols, C. H., & Newport, E. L. (1992). The role of stress and position in determining first words. Language Acquisition, 2, 189– 220.Fernald, A. (1985). Four-month-old infants prefer to listen to motherese. Infant Behavior and Development, 8, 181–195.Fernald, A. (1989). Intonation and communicative intent in mother’s speech to infants: Is the melody the message? Child Development, 60, 1497–1510.Fernald, A. (1993). Approval and disapproval: Infant responsiveness to vocal affect in familiar and unfamiliar languages. Child Development, 64, 657–674.Fernald, A., & Kuhl, P. K. (1987). Acoustic determinants of infant preference for parentese speech. Infant Behavior and Development, 10, 279–293.Fisher, C. (2000, July). Who’s blicking whom? Word order in early verb learning. Poster session presented at the 11th International Conference on Infant Studies, Brighton, England.Fisher, C. (2002). The role of abstract syntactic knowledge in language acquisition: A reply to Tomasello (2000). Cognition, 82, 259– 278.Gentner, D., Holyoak, K., & Kokinov, B. (Eds.). (2001). The analogical mind. Cambridge, MA: MIT Press.Givón, T. (1993). English grammar: A function-based introduction. Amsterdam: Benjamins.Gleitman, L. R., & Newport, E. L. (1995). The invention of language by children: Environmen- tal and biological influences on the acquisition of language. In L. R. Gleitman & M. Liberman (Eds.), An invitation to cognitive science: Vol. 2. Language (2nd ed., pp. 1–24). Cambridge, MA: MIT Press.Goldin-Meadow, S. (1985). Language development under atypical learning conditions: Replication and implications of a study of deaf children of hearing parents. In K. Nelson (Ed.), Children’s language: Vol. 5 (pp. 197–245). Hillsdale, NJ: Lawrence Erlbaum Associates.Goldin-Meadow, S. (2003). The resilience of language: What gesture creation in deaf children can tell us about how all children learn language. New York: Psychology Press.Goldstein, M., King, A., & West, M. (2003). Social interaction shapes babbling: Testing parallels between birdsong and speech. Proceedings of the National Academy of Sciences, 100, 8030– 8035.Halliday, M. A. K. (1975). Learning how to mean. London: Edward Arnold.Harris, M., Jones, D., Brookes, S., & Grant, J. (1986). Relations between the non-verbal context of maternal speech and rate of language development. British Journal of Developmental Psychol- ogy, 4, 261– 268.Harris, M., Jones, D., & Grant, J. (1983). The nonverbal context of mothers’ speech to infants. First Language, 4, 21–30.Hirsh-Pasek, K., Golinkoff, R., & Naigles, L. (1996). Young children’s use of syntactic frames to[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... derive meaning. In K. Hirsh-Pasek & R. Golinkoff (Eds.), The origins of grammar (pp. 123–158). Cambridge, MA: MIT Press.Hoff, E., & Naigles, L. (2002). How children use input to acquire a lexicon. Child Development, 73, 418–433.Hollich, G. J., Hirsh-Pasek, K., & Golinkoff, R. M. (2000). Breaking the language barrier: An emergentist coalition model for the origins of word learning. Monographs of the Society for Research in Child Development, 65 (3, Serial No. 262). 114. 104 Dare Baldwin and Meredith MeyerHyams, N. (1986). Language acquisition and the theory of parameters. Dordrecht, The Netherlands: Reidel.Jaswal, V. K. (2004). Don’t believe everything you hear: Preschoolers’ sensitivity to speaker intent in category induction. Child Development, 75, 1871–1885.Johnson, S. C. (2000). The recognition of mentalistic agents in infancy. Trends in Cognitive Sciences, 4, 22–28.Jusczyk, P. (1997). The discovery of spoken language. Cambridge, MA: MIT Press.Kaye, K. (1982). The mental and social life of babies: How parents create persons. Chicago: University of Chicago Press.Klin, A. (1991). Young autistic children’s listening preferences in regard to speech: a possible characterization of the symptom of social withdrawal. Journal of Autism and Developmental Disorders, 21, 29– 42.Klin, A. (1992). Listening preferences in regard to speech in four children with developmental disabilities. Journal of Child Psychology and Psychiatry, 33, 763–776.Kuhl, P. K. (2004). Early language acquisition: Cracking the speech code. Nature Reviews Neu- roscience, 5, 831–843.Kuhl, P. K., Andruski, J. E., Chistovich, I. A., Chistovich, L. A., Kozhevnikova, E. V., Ryskina, V. L., et al. (1997). Cross language analysis of phonetic units in language addressed to infants. Science, 277, 684–686.Kuhl, P. K., Coffey- Corina, S., Padden, D., & Dawson, G. (2005). Links between social and linguistic processing of speech in preschool children with autism: Behavioral and electrophysi- ological evidence. Developmental Science, 8, F1–F12.Kuhl, P. K., Tsao, F.-M., & Liu, H.-M. (2003). Foreign-language experience in infancy: Effects of short-term exposure and social interaction on phonetic learning. Proceedings of the National Academy of Sciences, 100, 9096– 9101.Landau, B., & Gleitman, L. R. (1985). Language and experience: Evidence from the blind child. Cambridge, MA: MIT Press.Langacker, R. W. (1987). Foundations of cognitive grammar: Vol. 1. Stanford, CA: Stanford University Press.Lidz, J., & Waxman, S. (2004). Reaffirming the poverty of the stimulus argument: a reply to the replies. Cognition, 93, 157–165.Lidz, J., Waxman, S., & Freedman, J. (2003). What infants know about syntax but couldn’t have learned: Evidence for syntactic structure at 18 months. Cognition, 89, B65–B73.Liu, H.-M., Kuhl, P. K., & Tsao, F.-M. (2003). An association between mothers’ speech clarity and infants’ speech discrimination skills. Developmental Science, 6, F1–F10.Lyons, J. (1977). Semantics: Vol. 1. Cambridge: Cambridge University Press.Marcus, G. F. (1993). Negative evidence in language acquisition. Cognition, 46, 53– 85.Markman, E. M. (1989). The development of categories and category names: Problems of induction. Cambridge, MA: MIT Press.Moore, C.,[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Angelopoulos, M., & Bennett, P. (1999). Word learning in the context of referential and salience cues. Developmental Psychology, 35, 60–68.Moses, L. J., & Flavell, J. H. (1990). Inferring false beliefs from actions and reactions. Child Development, 61, 929–945.Mundy, P., Sigman, M., Ungerer, J., & Sherman, T. (1986). Defining the social deficits of autism: The contribution of nonverbal communication measures. Journal of Child Psychology and Psychiatry, 27, 657–669.Naigles, L. (1990). Children use syntax to learn verb meaning. Journal of Child Language, 17, 357–374. 115. How Inherently Social is Language? 105Nazzi, T., Bertoncini, J., & Mehler, J. (1998). Language discrimination by newborns: Toward an understanding of the role of rhythm. Journal of Experimental Psychology: Human Perception and Performance, 24, 756–766.Nelson, K. (1985). Making sense: The acquisition of shared meaning. New York: Academic Press.Newport, E., Gleitman, H., & Gleitman, L. R. (1977). Mother, I’d rather do it myself: Some effects and noneffects of maternal speech style. In C. E. Snow & C. A. Ferguson (Eds.), Talking to children: Language input and acquisition (pp. 109–150). Cambridge: Cambridge University Press.Osterling, J., Dawson, G., & Munson, J. (2002). Early recognition of 1-year-old infants with autism spectrum disorder versus mental retardation. Development and Psychopathology, 14, 239–251.Perner, J., & Ruffman, T. (2005). Infants’ insight into the mind: How deep? Science, 308, 214–216.Pinker, S. (1994). The language instinct. London: Allen Lane.Povinelli, D. J., & Vonk, J. (2003). Chimpanzee minds: Suspiciously human? Trends in Cognitive Sciences, 7, 157–160.Preissler, M. A., & Carey, S. (2005). The role of inferences about referential intent in word learning: Evidence from autism. Cognition, 97, B13– B23.Quine, W. V. O. (1960). Word and object. Cambridge, MA: MIT Press.Ruffman, T. (2003). Does the infant have a theory of mind? New Zealand: University of Otago.Sabbagh, M., & Baldwin, D. A. (2005). Understanding the role of communicative intentions in word learning. In N. Eilan, C. Hoerl, T. McCormack, & J. Roessler (Eds.), Joint attention: Communication and others’ minds (pp. 165–184). Oxford: Oxford University Press.Saffran, J. R., Aslin, R. N., & Newport, E. L. (1996). Statistical learning by 8-month-old infants. Science, 274, 1926–1928.Samuelson, L. K., & Smith, L. B. (1998). Memory and attention make smart word learning: An alternative account of Akhtar, Carpenter, and Tomasello. Child Development, 69, 94– 104.Saylor, M., Sabbagh, M. A., & Baldwin, D. A. (2002). Children use whole- part juxtaposition as a pragmatic cue to word meaning. Developmental Psychology, 38, 993–1003.Saylor, M. M., Baldwin, D. A., & Sabbagh, M. A. (2004). Converging on word meaning. In D. G. Hall & S. R. Waxman (Eds.), Weaving a lexicon (pp. 509–531). Cambridge, MA: MIT Press.Shatz, M. (1981). Learning the rules of the game: Four views of the relation between grammar acquisition and social interaction. In W. Deutsch (Ed.), The child’s construction of language (pp. 17–38). London: Academic Press.Shatz, M. (1983). Communication. In J. H. Flavell & E. M. Markman (Eds.), Handbook of child psychology: Vol. 3. Cognitive development (3rd ed., pp. 841–890).[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... New York: Wiley.Shatz, M. (1992). A forward or backward step in the search for an adequate theory of language acquisition. Social Development, 1, 151– 154.Shatz, M. (1994). Theory of mind and the development of social-linguistic intelligence in early childhood. In C. Lewis & P. Mitchell (Eds.), Children’s early understanding of mind: origins and development (pp. 311–329). Hillsdale, NJ: Lawrence Erlbaum Associates.Shimizu, Y. A., & Johnson, S. C. (2004). Infants’ attribution of a goal to a morphologically unfamiliar agent. Developmental Science, 7, 425–430.Slobin, D. I. (1982). Universal and particular in the acquisition of language. In E. Wanner & L. R. Gleitman (Eds.), Language acquisition: The state of the art (pp. 128–170). Cambridge: Cambridge University Press. 116. 106 Dare Baldwin and Meredith MeyerSnow, C. E., & Ferguson, C. A. (Eds.). (1977). Talking to children: Language input and acquisi- tion. Cambridge: Cambridge University Press.Stern, D. N., Speiker, S., & MacKain, K. (1982). Intonational contours as signals in maternal speech to prelinguistic infants. Developmental Psychology, 18, 727–735.Tamis-LeMonda, C. S., Bornstein, M. H., Baumwell, L., & Damast, A. M. (1996). Responsive parenting in the second year: Specific influences on children’s language and play. In C. S. Tamis-LeMonda (Ed.), Parenting sensitivity: Individual, contextual and cultural factors in recent conceptualizations: Thematic issue of Early Development and Parenting, 5, 173–183.Thiessen, E. D., Hill, E., & Saffran, J. R. (2005). Infant-directed speech facilitates word seg- mentation. Infancy, 7, 53–71.Tomasello, M. (1992). First verbs: A case study in early grammatical development. Cambridge: Cambridge University Press.Tomasello, M. (1995). Language is not an instinct. Cognitive Development, 10, 131–156.Tomasello, M. (1999). The cultural origins of human cognition. Cambridge, MA: Harvard University Press.Tomasello, M. (2000a). Perceiving intentions and learning words in the second year of life. In M. Bowerman & S. Levinson (Eds.), Language acquisition and conceptual development (pp. 132–158). Cambridge: Cambridge University Press.Tomasello, M. (2000b). Do young children have adult syntactic competence? Cognition, 74, 209–253.Tomasello, M. (2003). Constructing a language. Cambridge, MA: Harvard University Press.Tomasello, M. (2004). Syntax or semantics? Response to Lidz et al. Cognition, 93, 139– 140.Tomasello, M., & Farrar, J. (1986). Joint attention and early language. Child Development, 57, 1454–1463.Tomasello, M., & Todd, J. (1983). Joint attention and lexical acquisition style. First Language, 4, 197–212.Van Valin, R. (1993). A synopsis of role and reference grammar. In R. Van Valin (Ed.), Advances in role and reference grammar (pp. 1–164). Amsterdam: Benjamins.Waxman, S. R. (2003). Links between object categorization and naming: Origins and emergence in human infants. In D. H. Rakison & L. M. Oakes (Eds.), Early category and concept develop- ment: Making sense of the blooming buzzing confusion (pp. 213–241). New York: Oxford University Press.Wellman, H. J. (1990). The child’s theory of mind. Cambridge, MA: MIT Press.Werker, J. F., & McLeod, P. J. (1989). Infant preference for both male[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... and female infant-directed talk: A developmental study of attentional and affective responsiveness. Canadian Journal of Psychology, 43, 230– 246.Wimmer, H., & Perner, J. (1983). Beliefs about beliefs: representation and constraining function of wrong beliefs in young children’s understanding of deception. Cognition, 13, 103–128.Woodward, A. L. (2003). Infants’ developing understanding of the link between looker and object. Developmental Science, 6, 297–311.Woodward, A. L., & Guajardo, J. J. (2002). Infants’ understanding of the point gesture as an object-directed action. Cognitive Development, 83, 1–24. 117. 6Input and the Acquisition of Language:Three QuestionsVirginia C. Mueller Gathercole and Erika HoffWhat is the role of input in the language acquisition process? Obviously, infants spokento in a given language reliably become children who speak that language, demonstratingin a general way that input must affect language development. But questions concerningthe role of input go beyond this obvious level and lie ultimately at the heart of the lan-guage acquisition process itself. Three central questions are: (1) What is the nature ofthe input, and what information about the grammar can the child extract from it? (2)Does input control either the sequence in which or the speed with which children con-struct the grammar? (3) Is the input alone sufficient to explain the child’s constructionof the grammar, or do other factors contribute to the process of acquisition; if so, howdo these interact with the input? The aim of this chapter is to review the theoreticalpositions on these questions and to examine the available evidence. We focus on the roleof input in the acquisition of language structure, the subject of the most long- standingand vigorous debates.Q1: What is the nature of the input, and what information aboutthe grammar can the child extract from it?Theories of the role of inputThe nativist view: Input plays a minor role. One of the staunchest positions on the inputis that taken by many nativists. In response to the three questions above they have argued(1) that the input to the child is an inadequate database from which to induce languagestructure, (2) that children need relatively little exposure to the input to induce the 118. 108 Virginia C. Mueller Gathercole and Erika Hoffstructure of the language, and input has little to do with sequence or speed of acquisi-tion, and (3) that children must be attributed with innate linguistic knowledge for themto be able to construct language. The nativist position is grounded in Chomsky’s (e.g., 1965, 1968, 1975) descriptionof language as a system of marvelous complexity, his assertion that a description of thatsystem is a description of linguistic knowledge represented in the human mind, and thecorollary assertion that studying the acquisition of language is thus to study how thelanguage-specific system “flowers” from that knowledge. With this, Chomsky alsoclaimed that children acquire language “on relatively slight exposure and without specifictraining” (Chomsky, 1975, p. 4). Furthermore, he argued, the input could not be veryimportant because it is an inadequate database from which to induce language structure.This “poverty of the[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... stimulus” assertion has two component claims: (1) that the speechchildren hear is full of errors, and (2) that any set of sentences in a language is, in prin-ciple, inadequate as a database because the underlying structure of language is not fullyrevealed in surface structures of sentences. Chomsky also asserted that general-purposelearning mechanisms operating on input alone would be insufficient to construct thegrammar of any language. These claims that the knowledge acquired is complex, thatthe available data are insufficient, and that the learning mechanisms are inadequatetogether have been termed “the logical problem of language acquisition” (Baker &McCarthy, 1981). The nativist solution to this problem has been to attribute innate lin- guistic knowledge of the universal properties of language to the child. That universalknowledge is then said to guide the child in constructing the language-particular instan-tiation of those universals from the input (see also Lidz, this volume). Since the original formulation of this problem, proposals concerning exactly whatis innate and how children manage to learn the particulars of the language they hearhave been refined (see, e.g., Crain & Thornton, 1998; Pinker, 1994; discussions inMacWhinney, 2004, and Sabbagh & Gelman, 2000, and commentaries). Among theproposals is the parameter setting model of acquisition (e.g., Hyams, 1986; Roeper &Williams, 1987), which attributes complex sets of parameters to the innate endowmentof the child. Each parameter may give the child a choice of two or three “settings,” andthe child’s job as an acquirer of the language is to determine from the input which settingfits the language s/he is hearing. (For example, the “pro-drop” parameter specifies thata language can have either obligatory overt subjects, like English (he was walking), oroptional overt subjects, like Spanish (_caminaba).) Determining the correct parametersetting might be complicated because it may involve several correlated features of thegrammar. (For example, whether or not a language allows pro-drop is correlated withwhether that language allows expletive subjects (as in it is raining), or has “real” auxil-iaries (may, can), without person, tense, and number marking; see Hyams, 1987.) Criti-cally, the theory explicitly holds that the innate parameters are designed in such a waythat the child can set each parameter on the basis of very minimal information in theinput, according to a “subset principle” (Berwick, 1985; Wexler & Manzini, 1987; butsee Atkinson, 2001; Lust, 1999). The role of the input is simply to act as a “trigger” forsetting parameters. This view has engendered many debates and proposed alternatives(see Goodluck, this volume; Drozd, 2004; Sabbagh & Gelman, 2000; and commentar-ies for recent discussions). 119. Input and the Acquisition of Language 109 It may even be possible, under the nativist position, to acquire language in the absenceof input. Nativists point to the development of fully complex creoles from grammaticallysimpler pidgins as children acquire pidgins as their native language (see, e.g., Bickerton’s(1981, 1984) bioprogram hypothesis). A recent case in point is the development of theIdioma de Señas Nicaragüense/Nicaraguan Sign Language (ISN). Kegl and colleagueshave[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... documented the rise of ISN from the 1970s, when a Nicaraguan school for the deafwas opened (Kegl, 2002, 2004; Kegl, Senghas, & Coppola, 1999; Senghas, Kita, &Özyürek, 2004). This new full language grew out of disparate – and very basic – “homesign” and gesturing systems used in individual families before the families’ comingtogether in the school. Proponents of the nativist position argue that the complexificationof the gesturing systems into ISN occurred within a single generation and was possiblebecause the children learning the sign system as their native language contributed aspectsof their innate linguistic knowledge to develop a more abstract, more complex system.Some have counter-argued, however, that the creolization process does not reflect thecontribution of Universal Grammar to pidgins but inter-borrowing of linguistic patternsfrom the native languages of the adults into the creoles children create (e.g., Goodman,1985; Maratsos, 1984; Lightfoot, 1984). Furthermore, the complexification process inthe case of both oral creoles and ISN may be a result of shortcuts typical of grammati-cization (Slobin, 1997), which is also not necessarily dependent on innate knowledge. Itis of note as well that the development of ISN occurred over more than one generation(Senghas & Coppola, 2001), which one might argue is counter to the expectation if thecomplexification arose out of the individual children’s access to Universal Grammar.Alternative views: Input plays a major role. The nativist position has been challengedon a number of general grounds. Alternative linguistic theories have challenged theChomskyan position on the nature of adult grammar. Cognitive and functionalist theo-ries ground language structure in general properties of human cognition and in thecommunicative functions of language (Culicover & Jackendoff, 2005; Foley & VanValin, 1984; Tomasello, 1995, 2003). Theories of acquisition based on these descriptionsof the grammar argue that children achieve grammar via its basis in communicativefunction (Bates & MacWhinney, 1989; Budwig, 1995). Construction grammars (Croft,2001; Goldberg, 1995) posit that grammars consist of networks of constructions, basedto a degree on meaning and existing at multiple levels of concreteness and abstraction.Language, in these alternative views, is not less complex, but it is less abstract than inthe Chomskyan descriptions. In addition, universals of language are posited to lie notin innate linguistic structures, but in universal cognitive structures and universalsof the human condition (Croft, 2001; Tomasello, 1995, 2003). Under these theories,language acquisition is more plausibly achievable without innate language-specificknowledge. The less abstract constructs posited make language more accessible throughthe input, and the child’s task can be taken as one of induction from the input(MacWhinney, 2004). Other challenges to Chomskyan nativism have focused more directly on the role ofinput and have argued against the claims that input is deficient and that children relyonly minimally on input to construct a grammar. This work is of two sorts: (1) illustra-tions that the input is more well-formed and revealing of linguistic structure than 120. 110 Virginia C. Mueller Gathercole and Erika Hoffnativists had argued,[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... and (2) evidence that patterns in the input are associated withpatterns in children’s developing language, suggesting that language acquisition makesdirect use of distributional patterns in the input. We will examine these in turn.Descriptions of the inputMotherese. In response to the claim that input is deficient, early research first took acloser look at the nature of the input. The initial work asked whether, in fact, input tochildren is errorful and therefore a deficient database from which to derive the regulari-ties of language. The clear finding was that, when talking to children, adults producespeech that is slow and highly grammatical and that has a higher pitch and broader pitchrange than speech among adults (Fernald et al., 1989; see Gerken, 1994). Furthermore,adults adjust the complexity of their speech, at least grossly, to the child’s level of com-prehension (Snow & Ferguson, 1977). Beyond simplifying their speech, adults also tendto follow the child’s attentional focus, produce multiple utterances on the same topic,ask questions, and provide contingent replies; these may have their own consequencesfor language learning beyond those posited as contingent on the simplification processes.This special register for talking to children was dubbed “motherese” (Newport,Gleitman, & Gleitman, 1977). Subsequent work revealed that the high pitch and exag-gerated intonation contour of motherese made it especially interesting to infants (Cooper& Aslin, 1994; Fernald, 1985). One hypothesis was that the correspondence betweenintonation contour and grammatical structure might make this special register helpfulto children’s learning of language structure; this was supported by the finding thatinfants preferred to listen to exaggerated contours that corresponded to phrase boundar-ies over equally varied patterns that did not (Hirsh-Pasek et al., 1987). Another suggestion was that motherese supported language development by providinga simpler model of language than does adult-directed speech and, by extension, thatwithin the variability in child- directed speech that exists, simpler is better. That latterhypothesis finds little support in the evidence. There is one finding in the literaturethat shorter maternal mean lengths of utterance are positively related to children’ssyntactic development (Furrow, Nelson, & Benedict, 1979), but that finding has neverbeen replicated despite multiple attempts to do so (Pine, 1994). To the contrary,several studies have found that children who hear longer utterances in input are moreadvanced in syntactic development (Harkness, 1977; Hoff- Ginsberg, 1998; Huttenlo-cher, Vasilyeva, Cymerman, & Levine, 2002). Additionally, some input features that arepositively associated with children’s syntactic development, such as adult question-asking,involve grammatically complex forms. (Despite such findings, it may still be the casethat the average degree of simplification in child-directed speech benefits languageacquisition. All of the observed benefits of complexity in mothers’ speech have beenobtained within a range of complexity that was more limited than in speech directed toadults.) One recent hypothesis has suggested that the child, not the caregiver, may be thesource of simplification of input. Newport (1990) and her colleagues have suggested that[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... 121. Input and the Acquisition of Language 111the limited perceptual and memory capacities of young children give the child an advan-tage. If the child can process and store only “pieces” of the input, this facilitates analysis,because it minimizes the logically possible combinatorial hypotheses the child will haveto consider. Thus, “less is more.” The child will access more complex forms when readyto take in larger chunks of input. Another argument against a critical role for motherese draws on the considerablevariation that exists across cultures in the extent to which parents modify their speechto children, or even speak directly to children (Lieven, 1994). Despite a wide range inpatterns (e.g., Ochs, 1985; Schieffelin, 1985), children still learn language. It is argued,therefore, that because motherese is not universal, language development cannot becontingent on the child hearing motherese (see Hoff, 2006).Input as a source of corrective feedback? A second potential characteristic of input thatresearch addressed early on is the provision of corrective feedback for error. If inputprovided corrective feedback, this would contradict Chomsky’s claim that childrenreceive no training in language. An early study found that mothers did not correct theirchildren’s ungrammatical utterances (Brown & Hanlon, 1970). Furthermore, childrenseem remarkably resistant even when parents do make occasional corrections. On thebasis of these findings, the consensus in the field has long been that children do notgenerally receive corrective feedback. More recent work has explored whether adults may provide more subtle feedback.Some have found that when children produce well-formed utterances, adults are morelikely to repeat them verbatim, whereas when children produce ungrammatical forms,adults are more likely to modify them, to provide correct forms, or to ask for clarification(Bohannon & Stanowicz, 1988; Chouinard & Clark, 2003; Demetras, Post, & Snow,1986; Saxton, 1997; Saxton, Backley, & Gallaway, 2005). Just how useful this feedbackis to the child is a matter of debate. Chouinard and Clark (2003) argue that childrendo frequently recognize reformulations as corrections; Saxton et al. (2005) reportthat contrastive use of correct forms by adults predicts changes in children’s errorrates. Countering this view, Atkinson (2001) notes that demonstrations of occasionalfeedback do not necessarily mean that the adult reliably signals the grammaticality ofchildren’s utterances, nor that such feedback is a necessary element of acquisition. Giventhe probabilistic nature of feedback, Marcus (1993) estimated that a child would haveto say the same ungrammatical sentence 85 times in order to have enough data to deter-mine that the sentence was ungrammatical. Any feedback the child is receiving, there-fore, can be seen, at best, only as an aid to language development; language developmentcannot be seen as contingent on such feedback. Moreover, Shatz and Ebeling (1991)argued that children actually revise their own utterances syntactically more than theirparents do.Input as data for distributional learning. The theoretical importance of feedback declinesif structural properties of the language can be induced directly from distributional pat-terns in the input, as argued first by Maratsos and Chalkley[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... (1980). Several sources of 122. 112 Virginia C. Mueller Gathercole and Erika Hoffevidence support this possibility, ranging from computer simulations of language devel-opment, to analyses of distributions of forms in parental input, to evidence of children’sattention to frequency distributions in the input. Computational models have demonstrated clearly that computers can induce gram-matical features of language and syntax–semantics mappings. Redington and Chater(1997), for example, have shown that given a large sample of speech (including adult-to-adult and adult-to-child speech) as input, computer models can extract word classes(nouns, verbs, etc.), often posited as innate, from distributional patterns in that input.Smith and colleagues (Colunga & Smith, 2005; Gasser & Smith, 1998) have shownsimilarly that nouns can be distinguished from adjectives based on characteristics of theinput, and that the association of nouns with solids and non-solids can emerge throughassociative learning. Landauer and Dumais (1997) have demonstrated that a computercan “learn” semantic associations among words using simple associationist mechanisms.Such demonstrations confirm that, in principle, structure-relevant patterns are availablein the input language. In direct examinations of speech addressed to children, Mintz (2003) found thatfrequently occurring word frames reliably surround words of the same grammatical cat-egory, and Naigles and Hoff-Ginsberg (1995) found that verbs in different semanticcategories (e.g., internal state verbs, motion verbs) appeared in different syntactic envi-ronments. These findings make plausible the argument that children could induce formclasses from the input. Pine and his colleagues (Gobet, Freudenthal, & Pine, 2004) havegone one step further: in computer simulations of children’s language development usingreal parental speech to children as input, they have successfully simulated a number ofphenomena observed in children’s language, including the emergence of optional infini-tive phenomena, verb-island phenomena, subject omissions, and case marking errors. These studies examining patterns in the input and simulations of learning based oninput data are complemented by behavioral studies showing that children can and doextract patterns of language from the input. Recent research has provided an explosionof evidence that human infants are powerful learners, able to extract information aboutthe perceptual properties of language from the distributional properties of the speechthey hear. Jusczyk and colleagues (e.g., Jusczyk, 1997) have demonstrated that infantshave keen abilities to attend to stress, prosody, syllable, and lexical patterns in speech,and these aid in the infants’ extraction of patterns in the input. Saffran and colleagueshave found reliable effects showing that infants carry out rapid learning of statisticalprobabilities in language – whether those have to do with phonological patterns, lexicalitems, or phrase structure patterns (see, e.g., Saffran, 2003). These abilities to track sta-tistical probabilities are not limited to language but extend also to tones and visual pat-terns as well as to rhythmic patterns (Hannon & Trehub, 2005; Kirkham, Slemmer, &Johnson, 2002; Saffran & Thiessen, this volume).[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... However, infants do favor certainpatterns over others, in particular, patterns consistent with those found in the world’slanguages (Saffran, 2003). But, importantly, since these abilities at tracking statisticalprobabilities are present in other species as well, Saffran (2003) argues that “the similari-ties across languages . . . are not the result of innate linguistic knowledge. Instead,human languages have been shaped by human learning mechanisms” (p. 110). (See alsoGerken; Saffran & Thiessen, this volume.) 123. Input and the Acquisition of Language 113 A rich body of research with toddlers has suggested how this early sensitivity to formin the input provides the basis for language-specific morphosyntactic and semanticinductions. For example, English-speaking children learn early that a word followinga probably refers to an object, while the same new word following some (someblicket) probably refers to a substance (Bloom, 1994; Carey, 1994; Gathercole, Cramer,Somerville, & Jansen op de Haar, 1995; Gordon, 1988; Soja, 1992; Soja, Carey, &Spelke, 1991). English-speaking children also use the correspondences between thesemantics of verbs and the structures in which they appear as clues to verb meaning(Naigles & Hoff-Ginsberg, 1998; Naigles & Swensen, this volume). Similarly, English-speaking children who hear a new word ending in -ish (foppish) learn to infer that thatword refers to a property of an object, not the object itself, while Spanish-speaking chil-dren treat adjective-like forms in their language as if they refer to objects, because of thestructure of Spanish (Waxman, Senghas, & Benveniste, 1997). Welsh- speaking childrenlearn that if they hear a new word in a noun slot, that word might refer to either a singleobject or to a collection, while English- and Spanish-speaking children learn that a newnoun in their language is likely to refer to a single whole object (Gathercole, Thomas,& Evans, 2000). In a similar vein, Spanish-speaking children learn that a new verbreferring to a motion is likely to incorporate the direction of motion (as in Juan subió lacolina corriendo, “Juan ascended the hill running”), while English-speaking childrenlearn that a new verb referring to a motion is likely to incorporate the manner of motion(as in John ran up the hill) (Hohenstein, 2001). In addition, the distribution of forms in children’s speech often reflects the distribu-tional frequencies in the adult language, indicating that children are highly sensitive tothose distributions. Theakston, Lieven, Pine, and Rowland (2002) reported that chil-dren’s early use of the verb go in English involves several isolated structures linking syn-tactic form with semantics, and that these correspond highly with the structural formsof go in the input. De Villiers (1985) found that the structures in which young childrenused particular verbs corresponded to the structures in which their mothers used thoseverbs. Henry (2003) reported on children’s use of past forms in Ulster English, whichallows both irregular and regularized forms for many verbs. She found that the children’spatterns of usage closely followed those of the adults around them. Gobet et al. (2004)demonstrated that English-, Dutch-, and Spanish-speaking children’s use of bare infini-tives versus finite verb forms – which has figured prominently in nativist arguments(Wexler, 1994, 1998) – can be successfully[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... mimicked in computer simulations thatcombine simple distributional analyses with a well-known child strategy (Slobin, 1973)of focusing on the ends of input utterances. Together these studies support the position that the speech children hear containsstructure-revealing information and that children have the capacity to find that informa-tion and use it to induce structural properties of their language. Further, the evidenceargues that children are sensitive to the form–meaning correspondences of their languageand use these to predict and infer linguistic properties of new forms when they encounterthem (see also Choi, 2006; Gathercole, 2006; and see Lidz, this volume, on children’srecovery from incorrect inferences). Whether these abilities rely solely on the child’sattention to the input or involve other factors will be considered when we address ques-tion 3. We turn now to our second question. 124. 114 Virginia C. Mueller Gathercole and Erika HoffQ2: Does input control either the sequence in which or the speedwith which children construct the grammar?We begin with evidence indicating that input – or more properly, frequency of input –does not control the sequence in which forms are acquired (except in the limited case inwhich sufficient input is not available); we follow with evidence showing that input doesaffect speed of acquisition.Input and sequence of acquisitionIf the acquisition of language were a simple process of storing and mimicking the input,one might expect that the sequence in which forms develop would correspond directlywith their relative input frequency. It is clear that this is not the case. First, the formsthat often are the most frequent in the input, such as function words (like a, the, andof ), are rarely those that appear first in children’s speech. Second, if children simplycopied patterns available in the input, one would not expect them to make errors, atleast not frequently. Again, this is not the case. Third, examinations of developmentalsequences (e.g., for grammatical morphemes) make it clear that, assuming some minimalinput level, other factors such as linguistic complexity and perceptual salience (see below)are more influential in determining order of acquisition than frequency in the input(Brown, 1973; de Villiers & de Villiers, 1973). Even though input does not largely control the order in which children acquire theforms of language, input is relevant to order of acquisition in at least two ways. First,input affects order of acquisition in the extreme, in that children cannot acquire whatthey do not hear in the input. Several researchers have proposed, in fact, the need for a“critical mass” of input for acquisition or abstraction to occur (Conti-Ramsden & Jones,1997; Elman, 2003; Gathercole, 2002b, 2002c; Marchman & Bates, 1994). If such acritical mass is not available to a child, the relevant structure may not be acquired, maybe acquired late, or may not be acquired fully. Two cases in point are the passive in English and the present perfect in AmericanEnglish. English-speaking children learn the passive (e.g., he was beaten by his opponent)quite late. Some nativists have posited that this is due to the late maturation of certainrelevant innate linguistic principles (Borer & Wexler, 1987; see Goodluck, this volume).However, the passive is used[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... infrequently in English. Children learning languages inwhich the passive is more frequent have been observed to use the passive early (Allen &Crago, 1996; Demuth, 1989; Pye & Quixtan Poz, 1988). Furthermore, if the frequencyof passive forms is increased in speech to English-speaking children, children can learnsome aspects of the passive earlier (de Villiers, 1980). Another telling case is the acquisi-tion of the present perfect in American versus British English. While these two dialectsshare the same syntactic and semantic forms for the present perfect, American Englishuses the present perfect much less frequently than British English (optionally substitut-ing, for some uses of the present perfect, the regular past: Did you eat yet? for Have youeaten yet?). This difference in frequency affects timing of acquisition, with British chil- 125. Input and the Acquisition of Language 115dren using present perfect constructs by 3 years, and American children not until muchlater (Gathercole, 1986). The exact quantity that constitutes the “critical mass” for the acquisition of a structuremay be debatable, but appears to be linked with the relative transparency/opacity of thestructure. Structures that are transparent appear to require a lower critical mass thanopaque structures for abstraction of the relevant patterns. For example, grammaticalgender in Spanish, which is very transparent, is acquired at an early age (Cain, Weber-Olsen, & Smith, 1987; Hernández Pina, 1984); grammatical gender in Welsh, which isvery opaque, involving multiple overlapping form–function correspondences, is notlearned until after age 9 (Gathercole & Thomas, 2005; Gathercole, Thomas, & Laporte,2001; Thomas, 2001). Similarly, that-trace structures (e.g., ¿Quién piensas que tiene ojosverdes?/Who do you think (that omitted) has green eyes?) and the overall use of comple-mentizer que in Spanish are transparent and are learned early, whereas that-trace struc-tures and the overall use of complementizer that in English are opaque and learned late(Gathercole, 2002c; Gathercole & Montes, 1997). A second way in which frequency of input might affect sequence of acquisition, or,rather, the nature of acquisition in that sequence, has to do with the extent to whichchildren generalize beyond learned instances. If children construct language from pat- terns in the input, one theoretical question is how they arrive at appropriate levels ofproductivity for structures. Productivity is essential to linguistic knowledge, for it allowsthe use of language beyond learned instances. We know that at certain points in devel-opment children can extend their grammatical knowledge to novel forms (e.g., Berko,1958), and they generalize and overgeneralize to novel instances – for example, usingregularized forms in place of irregular items (e.g., falled instead of fell; I disappeared itinstead of I made it disappear). Two critical questions regarding (over)generalizationsconcern (1) when these forms occur and (2) how the child manages to eventually elimi-nate incorrect forms from his/her speech. Tomasello (2000) has recently argued thatinput frequency plays a role in determining when overgeneralizations occur in childspeech. He argues that specific items that are frequently heard become entrenched, andsuch items[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... are less likely to be overgeneralized than items that are less frequent in input.For example, the high-frequency verb laugh becomes entrenched as an intransitive verband is not likely to be overgeneralized to a transitive use, I laughed him. Overgeneraliza-tions are also constrained by presence in input of alternative forms, which preemptovergeneralization. In this way, hearing the construction made X disappear blocks theovergeneralization of disappear to a transitive use, I disappeared it (Brooks & Tomasello,1999). Items of an intermediate level of frequency (e.g., giggle) are the most susceptibleto overgeneralization (You giggled me), according to this argument, because the appropri-ate forms are not heard frequently enough to become entrenched, yet are likely to belearned before the system is fully worked out. There is empirical support for some pos-tulates of this account (Brooks, Tomasello, Dodson, & Lewis, 1999), although the issueof how children manage to be productive language users without being wildly overpro-ductive has not been fully resolved (see Maratsos, 2000, and Elman, 2003, for relatedarguments). Concerning the sequence in which constructs develop, then, frequency of input perse does not control order of acquisition. Something else does. However, frequency of 126. 116 Virginia C. Mueller Gathercole and Erika Hoffinput does affect the availability of a structure, and for each structure, the child mustaccumulate enough experience to be able to draw the relevant generalizations when readyto do so. If the structures in question are quite transparent, that critical mass will besmaller than if the structures in question are quite opaque. In addition, as the child isaccumulating that critical mass, s/he is not likely to make errors of overgeneralizationon entrenched items (learned early and heard frequently), only on items that are lessentrenched, and only before the system is fully worked out.Input and speed of acquisitionWhile relative frequency of input does not affect sequence of acquisition, it can influencespeed of acquisition. One “natural laboratory” source of evidence is in comparisons ofbilingual and monolingual groups learning the same pair of languages. Comparisons ofchildren growing up in contexts in which the relative proportion of input in languagesA and B varies yield consistent differences in the timing of acquisition of structures.Gathercole (2002a, 2002b, 2002c) found this to be the case for Spanish–English bilin-guals learning mass/count structures in English, grammatical gender in Spanish, andthat-trace structures in English and Spanish: Bilinguals who had the greatest amount ofEnglish input had an earlier command of the English constructs than their peers, whilebilinguals who had the greatest amount of Spanish input had an earlier command of theSpanish constructs than their peers. (This also meant that English and Spanish mono-linguals gained command of these structures before their bilingual peers.) Gathercoleand Thomas (2005) found this also to be the case for Welsh–English bilinguals learninggrammatical gender and verb-argument structures in Welsh: Those with a greater amountof Welsh input on a daily basis showed an earlier command of the Welsh constructsthan their peers with less Welsh input. (See[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... also Rieckborn (2006) and Kupisch (2003)for similar effects in bilinguals’ development of tense/aspect and determiners,respectively.) Monolingual children also differ in how much they hear the language or particularstructures in the language they are acquiring, and this affects their rate of grammaticaldevelopment. The total quantity of speech addressed to children at home and in daycare relates to children’s linguistic development (Bradley & Caldwell, 1976; Clarke-Stewart, 1973; McCartney, 1984; National Institute of Child Health and HumanDevelopment, 2000). The talkativeness of English- speaking mothers in interaction withyoung children relates to the children’s syntactic and semantic development (Barnes,Gutfreund, Satterly, & Wells, 1983). Kindergarten children whose teachers use morecomplex sentences grow more rapidly in their use of complex sentences than those whoseteachers produce fewer complex sentences (Huttenlocher et al., 2002). The more fre-quently children hear questions with auxiliary inversion, the more rapidly they grow intheir own use of auxiliary verbs (Hoff-Ginsberg, 1985; Newport et al., 1977; Shatz,Hoff-Ginsberg, & MacIver, 1989). The variety of syntactic frames in which childrenhear verbs used predicts the syntactic flexibility of children’s verb use (Naigles & Hoff-Ginsberg, 1998). The discourse environment of forms in input also affects languagedevelopment. Expansions and recasts by adults may positively predict syntactic develop- 127. Input and the Acquisition of Language 117ment (Newport et al., 1977), as may mothers’ inexact self-repetitions (Cross, 1978;Hoff-Ginsberg, 1985, 1986). Besides amount of input, other properties of children’s conversational experience havealso been shown to affect rates of grammatical development, including the amount oftime spent in joint attention (Carpenter, Nagell, & Tomasello, 1998; Laakso, Poikkeus,Katajamaki, & Lyytinen, 1999; Mundy & Gomes, 1998), maternal responsivity to childverbalizations (Tamis-LeMonda, Bornstein, & Baumwell, 2001), and contingency ofmaternal speech (Snow, Perlmann, & Nathan, 1987). The benefits of these features ofconversational experience might in the end reflect the amount of input provided thechild. That is, when mothers and children are more engaged in conversation, childrenreceive more language-advancing data. It may also be, however, that having a responsiveconversational partner motivates language acquisition by demonstrating to children thatcommunication is both possible and interesting (Hoff, 2003, 2006); alternatively, thepartner who shares focus with the child may be more likely to provide input in line withthe child’s cognitive abilities (see below). The upshot of all of this research is that (a) frequency of input per se does not controlthe sequence of acquisition across forms, but (b) input does affect rapidity of acquisition.More input means more rapid development – through a course of development whosesequence appears to be largely dictated by other factors. Why might the quantity of input affect the rapidity of acquisition of forms inlanguage? There are a number of possibilities:1 With more input, there is greater frequency of the tokens of any form, which may contribute to the better storage and retention of tokens (e.g., walked heard 30 times is more likely to be retained than walked heard twice). Token frequency[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... may be particularly important for the acquisition of isolated irregular forms (drank, flew) (Maratsos, 2000).2 More input may also entail more distinct contexts (linguistic and non-linguistic) in which tokens are heard. This may facilitate the mapping problem of a form with its sense.3 Greater input frequency likely entails greater frequency of lexical types participating in a given morphological or syntactic structure (e.g., hearing not only walked, but also talked, laughed, etc.). Frequency of types provides the “grist” for the language development mill that will help the child to construct morphosyntactic structures (here, use of -ed for past tense).4 Greater input frequency is likely to provide richer information on relations across tokens and types, thus enabling a faster and stronger construction of networks of forms in the child’s repertoire. That is, not only will the child be hearing each token (talked, flew) more often and in more non-linguistic contexts (e.g., flew in relation to a bird at the pond last week, in relation to a bug that has just flown in the window, etc.), but s/he will be hearing other types used with similar morphological forms (walked, laughed, threw, drew) in similar non-linguistic contexts (in reference to past time, in reference to time immediately preceding the utterance, etc.). It is likely that all of these factors contribute to success in the child’s ability to con-struct the language being learned from the available input. They constitute the elements 128. 118 Virginia C. Mueller Gathercole and Erika Hoffthat will make up the “critical mass” of data that will eventually allow the child toabstract out the common structures that link them.Q3: Is the input alone sufficient to explain the child’s constructionof the grammar, or do other factors contribute to the process ofacquisition; if so, how do these interact with the input?Let us turn now to the third question, regarding what might be needed in addition toinput for the child to construct a language. We have already seen above that input alonecannot explain the order of acquisition across structures. So what contributes to acquisi-tion besides input to explain the order?Influences other than inputA strong nativist position might explain the sequence of development across structuresas controlled by the innate Universal Grammar. Some have posited that innate knowl-edge comes on line according to a maturational program, which controls the sequenceof development and helps explain why children’s knowledge does not necessarily matchwhat might predominate in the input. Thus, for example, the late acquisition of func-tional categories (e.g., determiners and prepositions) relative to lexical categories (e.g.,nouns and verbs), or the later acquisition of tense marking relative to person marking,might well be explained according to different maturational schedules for distinct ele-ments of Universal Grammar (e.g., Grinstead, 2000; Radford, 1990, 1996). However,this is not the only possibility. Another possibility, one that does not rely on innateknowledge, is that what children can take from the input is dependent on their own“readiness” for attending to, noticing, or understanding what the input has to offer. Thatreadiness might be, at least in part, in the form of cognitive understanding or of thechild’s linguistic development up to that point.Cognitive understanding and language[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... acquisition. Input clearly interacts with cognitionin determining what the child acquires when. Some demonstrations of the role of cogni-tive preparedness come from the acquisition of lexical items (see, e.g., Rice’s (1980) classicstudy of children’s acquisition of color terms), but grammatical examples, the focus ofthis chapter, are also in evidence. Some of these come from children’s early misuses ofterms that involve, in adult usage, complex cognitive knowledge. For example, children’searly uses of comparative forms (which in their adult-like use demand some understand-ing of seriation or scalarity) often involve incorrect applications in contexts where inten-sification (“very X”), a simpler concept, would be appropriate (Gathercole, 1983); earlyuses of relative clauses may be for compounding instead of relativization (Tavakolian,1981). In a recent study, Shirai and Miyata (2006) demonstrated that Japanese-speakingchildren use past tense morphology productively long before they use it appropriately fordeictic past reference. In children’s use of object labels, the greater difficulty of under- 129. Input and the Acquisition of Language 119standing functions related to substances than to shape seems to make children’s extensionof word categories based on material function harder (Gathercole & Whitfield, 2001).These examples show that if what is frequent in the input corresponds to complex cogni-tive concepts, the forms may be learned early with a simpler meaning (as in the case ofthe comparative in English or the past in Japanese), or may wait for the child’s cognitiveunderstanding to advance to a certain level (as in the case of word categories based onsubstance functions). This is not to say, however, that cognition always drives language acquisition; thereverse is also possible, that language can help “push” the child to attend to aspects ofreferents and to develop certain cognitive concepts earlier (e.g., Bowerman, 1996; Choi,2006; Gopnik & Choi, 1990). (However, the cognitive options open for such manipula-tion by language may be within a certain available cognitive range: Gathercole, 2006;McCune, 2006.) But ultimately, input alone cannot control order of acquisition becauseit must interact with, among other things, the child’s cognitive understanding of theworld to which language is referring.Linguistic complexity and language acquisition. The order in which children acquire formsalso depends to some extent on linguistic complexity. Take, for example, the acquisition ofthe third person singular form of verbs. In Spanish, this is the first finite form that becomesproductive; in English, it is a relatively late development. In Spanish, this form can beconsidered the unmarked, least complex, form of the verb, while in English, the thirdperson singular can be considered a marked, complex, form (Gathercole, Sebastián, &Soto, 1999, 2002). Similarly, children typically acquire simple sentences before complexones (involving more than one clause). Likewise, as noted above, constructions that involveopaque form–function mappings (e.g., Welsh grammatical gender, English that-trace) takelonger to acquire than similar constructions involving more transparent form–functionmappings (e.g., Spanish gender, Spanish que) (see Smoczynska, 1985). Another example,from Morgan, Barrière, and Woll[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... (2006), comes from the acquisition of British SignLanguage, in which agreement morphology is learned late; they attribute this to the dif-ficulty of segmenting the relevant signs into morphemes and to the complexity ofsemantically and syntactically conditioned agreement rules which must be mastered. It should be noted that complexity may depend not only on the structure in questionbut also on the relationships between the given construct and others in the linguisticsystem. Researchers have long noted that acquisition of a form may hinge on the prioracquisition of simpler related forms (e.g., Brown’s (1973) “law of cumulative complex-ity”), a notion that has returned recently in the shape of “construction conspiracies”(Abbot-Smith & Behrens, 2002; Morris, Cottrell, & Elman, 2000). Abbot-Smith andBehrens, for example, demonstrate that their German subject learned stative passivesbefore eventive passives because he had already acquired the “source constructions” forthe former. Similarly, Rice (1980) found that all of the children in her color study tookmuch longer to learn the first two-color- word contrast than to add a third color term totheir system once they had acquired the first contrast. Children clearly build on previ-ously acquired knowledge to move forward in the development of their linguistic system.The knowledge they have already acquired can ultimately serve to help them make moreefficient use of new, related information in the input. 130. 120 Virginia C. Mueller Gathercole and Erika HoffConclusionThe questions addressed in this chapter concern the degree to which the input childrenreceive can explain the course that language development follows. When the modernfield of child language began in the 1960s, the dominant linguistic view was that lan-guage was an innate faculty of the human mind and that the complex structure of lan-guage is only faintly evident in the surface forms; thus the child’s achievement of thatcomplex structure could only be explained by positing innate linguistic knowledge. Theevidence reviewed in this chapter suggests that the role of input in accounting for thefact of language acquisition is much greater than this early view allowed. Studies suggestthat input does more than faintly reveal language’s abstract structure, and that humaninfants and children have the capacity to induce language structure from surface regu-larities. We have seen that not only is an input-dependent account of language acquisitionplausible, but that there is strong support for its validity. Variation in the amount andnature of the input children receive correlates with variation in the rate at which theyacquire language. This evidence suggests that input provides the database for languageinduction. The evidence reviewed in this chapter also suggests, at the same time, that the lan-guage acquisition mechanism is not solely input driven. The sequence in which thestructures of language are acquired does not directly reflect the frequency with whichstructures occur in the input. Cognitive preparedness on the part of the child, and thelinguistic complexity of the forms to be acquired, also play key roles. The fact that chil-dren’s cognitive understanding can influence the acquisition of language structuresmakes the point that language is not a completely isolable domain,[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... and the influence oflinguistic complexity brings us full circle to input. Linguistically complex structures are,in part, those for which it is difficult to discern consistent patterns in the input. Thus,the effects of linguistic complexity serve to underline the fact that the child’s extractionof regularities in the input must ultimately play a key role in the final analysis of howlanguage acquisition takes place.ReferencesAbbot-Smith, K., & Behrens, H. (2002, July). Construction conspiracies in the acquisition of the German passive. Paper presented at the 10th Conference of the International Association for the Study of Child Language, Madison, WI.Allen, S. E. M., & Crago, M. B. (1996). Early passive acquisition in Inuktitut. Journal of Child Language, 23, 129–155.Atkinson, M. (2001). Learnability and the acquisition of syntax. In S. Bertolo (Ed.), Language acquisition and learnability (pp. 15– 80). Cambridge: Cambridge University Press.Baker, C. L., & McCarthy, J. J. (Eds.). (1981). The logical problem of language acquisition. Cambridge, MA: MIT Press.Barnes, S., Gutfreund, M., Satterly, D., & Wells, G. (1983). Characteristics of adult speech which predict children’s language development. Journal of Child Language, 10, 65–84. 131. Input and the Acquisition of Language 121Bates, E., & MacWhinney, B. (1989). Functionalism and the competition model. In B. MacWhinney & E. Bates (Eds.), The crosslinguistic study of sentence processing (pp. 3–73). Cambridge: Cambridge University Press.Berko, J. (1958). The child’s learning of English morphology. Word, 14, 50–77.Berwick, R. (1985). The acquisition of syntactic knowledge. Cambridge, MA: MIT Press.Bickerton, D. (1981). Roots of language. Ann Arbor, MI: Karoma.Bickerton, D. (1984). The language bioprogram hypothesis. Behavioral and Brain Sciences, 7, 173– 222.Bloom, P. (1994). Semantic competence as an explanation for some transitions in language development. In Y. Levy (Ed.), Other children, other languages: Theoretical issues in language development (pp. 41–75). Hillsdale, NJ: Erlbaum.Bohannon, J. S., & Stanowicz, L. (1988). The issue of negative evidence: Adult responses to children’s language errors. Developmental Psychology, 24, 684–689.Borer, H., & Wexler, K. (1987). The maturation of syntax. In T. Roeper & E. Williams (Eds.), Parameter setting (pp. 123–172). Dordrecht: D. Reidel Publishing Company.Bowerman, M. (1996). The origins of children’s spatial semantic categories: Cognitive versus linguistic determinants. In J. J. Gumperz & S. C. Levinson (Eds.), Rethinking linguistic relativ- ity (pp. 145–176). Cambridge: Cambridge University Press.Bradley, R. H., & Caldwell, B. M. (1976). The relation of infants’ home environments to mental test performance at fifty-four months: a follow-up study. Child Development, 47, 1172–1174.Brooks, P. J., & Tomasello, M. (1999). How children constrain their argument structure con- structions. Language, 75, 720–738.Brown, R. (1973). A first language: The early stages. Cambridge, MA: Harvard University Press.Brown, R., & Hanlon, C. (1970). Derivational complexity and order of acquisition in child speech. In R. Brown (Ed.), Psycholinguistics (pp. 155–207). New York: Free Press.Budwig, N. (1995). A developmental-functionalist approach to child language. Mahwah,[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... NJ: Erlbaum.Cain, J., Weber-Olsen, M., & Smith, R. (1987). Acquisition strategies in a first and second lan- guage: Are they the same? Journal of Child Language, 14, 333–352.Carey, S. (1994). Does learning a language require the child to reconceptualize the world? In L. Gleitman & B. Landau (Eds.), The acquisition of the lexicon (pp. 143–167). Cambridge, MA: MIT Press.Carpenter, M., Nagell, K., & Tomasello, M. (1998). Social cognition, joint attention, and com- municative competence from 9 to 15 months of age. Monographs of the Society for Research in Child Development, 63 (4, Serial No. 255).Choi, S. (2006). Influence of language-specific input on spatial cognition: Categories of contain- ment. First Language, 26, 207– 232.Chomsky, N. (1965). Aspects of the theory of syntax. Cambridge, MA: MIT Press.Chomsky, N. (1968). Language and mind. New York: Harcourt Brace & World, Inc.Chomsky, N. (1975). Reflections on language. New York: Pantheon Books.Chouinard, M. M., & Clark, E. V. (2003). Adult reformulations of child errors as negative evi- dence. Journal of Child Language, 30, 637– 669.Clarke-Stewart, K. A. (1973). Interactions between mothers and their young children: charac- teristics and consequences. Monographs of the Society for Research in Child Development, 38 (6–7, Serial No. 153).Colunga, E., & Smith, L. B. (2005). From the lexicon to expectations about kinds: A role for associative learning. Psychological Review, 112, 347– 382.Conti-Ramsden, G., & Jones, M. (1997). Verb use in specific language impairment. Journal of Speech, Language and Hearing Research, 40, 1298– 1313. 132. 122 Virginia C. Mueller Gathercole and Erika HoffCooper, R. P., & Aslin, R. N. (1994). Developmental differences in infant attention to the spectral properties of infant-directed speech. Child Development, 65, 1663– 1667.Crain, S., & Thornton, R. (1998). Investigations in Universal Grammar: A guide to experiments on the acquisition of syntax and semantics. Cambridge, MA: MIT Press.Croft, W. (2001). Radical construction grammar: Syntactic theory in typological perspective. Oxford: Oxford University Press.Cross, T. (1978). Mothers’ speech and its association with linguistic development in young chil- dren. In N. Waterson & C. Snow (Eds.), The development of communication (pp. 199–216). Chichester, England: Wiley.Culicover, P., & Jackendoff, R. (2005). Simpler syntax. Oxford: Oxford University Press.De Villiers, J. G. (1980). The process of rule learning in child speech: A new look. In K. E. Nelson (Ed.), Children’s language: Vol. 2 (pp. 1–44). New York: Gardner Press.De Villiers, J. G. (1985). Learning how to use verbs: lexical coding and the influence of the input. Journal of Child Language, 12, 587– 595.De Villiers, J. G., & de Villiers, P. A. (1973). A cross-sectional study of the acquisition of gram- matical morphemes in child speech. Journal of Psycholinguistic Research, 2, 267–278.Demetras, M. J., Post, K. N., & Snow, C. E. (1986). Feedback to first language learners: The role of repetitions and clarification questions. Journal of Child Language, 13, 275–292.Demuth, K. (1989). Maturation, continuity and the acquisition of Sesotho passive. Language, 65, 56–80.Drozd, K. (2004). Learnability and linguistic[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... performance. Journal of Child Language, 31, 431–457.Elman, J. (2003). Generalization from sparse input. Proceedings of the 38th Annual Meeting of the Chicago Linguistic Society.Fernald, A. (1985). Four-month-old infants prefer to listen to motherese. Infant Behavior and Development, 8, 181– 195.Fernald, A., Taeschner, T., Dunn, J., Papousek, M., De Boyson-Bardies, B., & Fukui, I. (1989). A cross-language study of prosodic modifications in mothers’ and fathers’ speech to preverbal infants. Journal of Child Language, 16, 477–501.Foley, W., & Van Valin, R. (1984). Functional syntax and universal grammar. Cambridge: Cambridge University Press.Furrow, D., Nelson, K., & Benedict, H. (1979). Mothers’ speech to children and syntactic devel- opment: some simple relationships. Journal of Child Language, 6, 423– 442.Gasser, M., & Smith, L. B. (1998). Learning nouns and adjectives: A connectionist account. In K. Plunkett (Ed.), Language acquisition and connectionism (pp. 269–306). Hove: Psychology Press.Gathercole, V. C. (1983). Haphazard examples, prototype theory, and the acquisition of compara- tives. First Language, 4, 169–196.Gathercole, V. C. (1986). The acquisition of the present perfect: Explaining differences in the speech of Scottish and American children. Journal of Child Language, 13, 537– 560.Gathercole, V. C. M. (2002a). Command of the mass/count distinction in bilingual and mono- lingual children: An English morphosyntactic distinction. In D. K. Oller & R. E. Eilers (Eds.), Language and literacy in bilingual children (pp. 175–206). Multilingual Matters.Gathercole, V. C. M. (2002b). Grammatical gender in bilingual and monolingual children: A Spanish morphosyntactic distinction. In D. K. Oller & R. E. Eilers (Eds.), Language and lit- eracy in bilingual children (pp. 207–219). Multilingual Matters.Gathercole, V. C. M. (2002c). Monolingual and bilingual acquisition: Learning different treat- ments of that-trace phenomena in English and Spanish. In D. K. Oller & R. E. Eilers (Eds.), Language and literacy in bilingual children (pp. 220–254). Multilingual Matters. 133. Input and the Acquisition of Language 123Gathercole, V. C. M. (2006). Introduction to Special Issue: Language-specific influences on acquisition and cognition. First Language, 26, 5–17.Gathercole, V. C. M., Cramer, L., Somerville, S., & Jansen op de Haar, M. (1995). Ontolo- gical categories and function: Acquisition of new names. Cognitive Development, 10, 225– 251.Gathercole, V. C. M., & Montes, C. (1997). That-trace effects in Spanish- and English-speaking monolinguals and bilinguals. In A. T. Pérez-Leroux & W. R. Glass (Eds.), Contemporary per- spectives on the acquisition of Spanish: Vol. 1. Developing grammars (pp. 75–95). Somerville, MA: Cascadilla Press.Gathercole, V. C. M., Sebastián, E., & Soto, P. (1999). The early acquisition of Spanish verbal morphology: Across-the-board or piecemeal knowledge? International Journal of Bilingualism, 3, 133– 182.Gathercole, V. C. M., Sebastián, E., & Soto, P. (2002). The emergence of linguistic Person in Spanish-speaking children. Language Learning, 52, 679–722.Gathercole, V. C. M., & Thomas, E. M. (2005). Minority language[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... survival: Input factors influencing the acquisition of Welsh. In J. Cohen, K. McAlister, K. Rolstad, & J. MacSwan (Eds.), ISB4: Proceedings of the 4th International Symposium on Bilingualism. Somerville, MA: Cascadilla Press.Gathercole, V. C. M., Thomas, E. M., & Evans, D. (2000). What’s in a noun? Welsh-, English-, and Spanish-speaking children see it differently. First Language, 20, 55–90.Gathercole, V. C. M., Thomas, E. M., & Laporte, N. (2001). The acquisition of grammatical gender in Welsh. Journal of Celtic Language Learning, 6, 53–87.Gathercole, V. C. M., & Whitfield, L. (2001). Function as a criterion for the extension of new words. Journal of Child Language, 28, 87–125.Gerken, L. (1994). Child phonology: Past research, present questions, future direction. In M. A. Gernsbacher (Ed.)., Handbook of psycholinguistics (pp. 781–820). New York: Academic Press.Gobet, F., Freudenthal, D., & Pine, J. M. (2004). Modelling syntactic development in a cross- linguistic context. In W. G. Sakas (Ed.), Proceedings of the First COLING Workshop on Psycho- Computational Models of Human Language Acquisition.Goldberg, A. (1995). Constructions: A Construction Grammar approach to argument structure. Chicago: University of Chicago Press.Goodman, M. (1985). Roots of languages (Review article). International Journal of American Linguistics, 51, 109–137.Gopnik, A., & Choi, S. (1990). Do linguistic differences lead to cognitive differences: A cross- linguistic study of semantic and cognitive development. First Language, 10, 199–215.Gordon, P. (1988). Count/mass category acquisition: Distributional distinctions in children’s speech. Journal of Child Language, 15, 109–128.Grinstead, J. (2000). Case, inflection and subject licensing in child Catalan and Spanish. Journal of Child Language, 27, 119–155.Hannon, E. E., & Trehub, S. E. (2005). Tuning in to musical rhythms: Infants learn more readily than adults. Proceedings of the National Academy of Sciences, 102, 12639– 12643.Harkness, S. (1977). Aspects of the social environment and first language acquisition in rural Africa. In C. E. Snow & C. A. Ferguson (Eds.), Talking to children: Language input and acquisition (pp. 309–318). Cambridge: Cambridge University Press.Henry, A. (2003, April). Acquiring past tenses from variable input. Keynote speech, Fifth Annual Gregynog Conference on Child Language, Powys, Wales.Hernández Pina, F. (1984). Teorías psicosociolingüísticas y su aplicación a la adquisición del español como lengua materna. Madrid: Siglo XXI de España Editores, S.A. 134. 124 Virginia C. Mueller Gathercole and Erika HoffHirsh-Pasek, K., Kemler Nelson, D. G., Jusczyk, P. W., Wright Cassidy, K., Druss, B., & Kennedy, L. (1987). Clauses are perceptual units for young infants. Cognition, 26, 269–286.Hoff, E. (2003). The specificity of environmental influence: Socioeconomic status affects early vocabulary development via maternal speech. Child Development, 74, 1368–1378.Hoff, E. (2006). How social contexts support and shape language development. Developmental Review, 26, 55–88.Hoff-Ginsberg, E. (1985). Some contributions of mothers’ speech to their children’s syntactic growth. Journal of Child Language, 12, 367– 386.Hoff-Ginsberg, E. (1986). Function and structure in maternal speech:[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Their relation to the child’s development of syntax. Developmental Psychology, 22, 155–163.Hoff-Ginsberg, E. (1998). The relation of birth order and socioeconomic status to children’s language experience and language development. Applied Psycholinguistics, 19, 603–629.Hohenstein, J. (2001). Motion event similarities in English- and Spanish-speaking children. Unpub- lished doctoral dissertation, Yale University.Huttenlocher, J., Vasilyeva, M., Cymerman, E., & Levine, S. (2002). Language input at home and at school: Relation to child syntax. Cognitive Psychology, 45, 337–374.Hyams, N. M. (1986). Language acquisition and the theory of parameters. Dordrecht: D. Reidel Publishing Company.Hyams, N. M. (1987). The theory of parameters and syntactic development. In T. Roeper & E. Williams (Eds.), Parameter setting (pp. 1–22). Dordrecht: D. Reidel Publishing Company.Jusczyk, P. W. (1997). The discovery of spoken language. Cambridge, MA: MIT Press.Kegl, J. (2002). Language emergence in a language-ready brain: Acquisition issues. In G. Morgan & B. Woll (Eds.), Language acquisition in signed languages (pp. 207–254). Cambridge: Cambridge University Press.Kegl, J. (2004). Language emergence in a language-ready brain: Acquisition issues. In L. Jenkins (Ed.), Biolinguistics and the evolution of language. Amsterdam: John Benjamins.Kegl, J., Senghas, A., & Coppola, M. (1999). Creation through contact: Sign language emergence and sign language change in Nicaragua. In M. DeGraff (Ed.), Comparative grammatical change: The intersection of language acquisition, creole genesis, and diachronic syntax (pp. 179– 237). Cambridge, MA: MIT Press.Kirkham, N. Z., Slemmer, J. A., & Johnson, S. P. (2002). Visual statistical learning in infancy: Evidence of a domain general learning mechanism. Cognition, 83, B35–B42.Kupisch, T. (2003). On the relation between input frequency and acquisition patterns from a cross-linguistic perspective. In J. van Kampen & S. Baauw (Eds.), Proceedings of GALA 2003. LOT Occasional Series.Laakso, M.-L., Poikkeus, A.-M., Katajamaki, J., & Lyytinen, P. (1999). Early intentional com- munication as a predictor of language development in young toddlers. First Language, 19, 207–231.Landauer, T. K., & Dumais, S. T. (1997). A solution to Plato’s problem: The latent semantic analysis theory of acquisition, induction, and representation of knowledge. Psychological Review, 104, 211– 240.Lieven, E. V. M. (1994). Cross-linguistic and cross-cultural aspects of language addressed to children. In C. Gallaway & B. J. Richards (Eds.), Input and interaction in language acquisition (pp. 56–73). Cambridge: Cambridge University Press.Lightfoot, D. (1984). The relative richness of triggers and the bioprogram. Behavioral and Brain Sciences, 7, 198–199. 135. Input and the Acquisition of Language 125Lust, B. (1999). Universal grammar: The strong continuity hypothesis in first language acquisi- tion. In W. C. Ritchie & T. K. Bhatia (Eds.), Handbook of child language acquisition (pp. 111–153). San Diego, CA: Academic Press.MacWhinney, B. (2004). A multiple process solution to the logical problem of language acquisi- tion. Journal of Child Language, 31, 963–968.Maratsos, M. (1984). How degenerate is the input to creoles and where do its biases come from?[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Behavioral and Brain Sciences, 7, 199–200.Maratsos, M. (2000). More overregularizations after all: New data and discussion on Marcus, Pinker, Ullman, Hollander, Rosen & Xu. Journal of Child Language, 27, 183– 212.Maratsos, M., & Chalkley, M. A. (1980). The internal language of children’s syntax: The onto- genesis and representation of syntactic categories. In K. Nelson (Ed.), Children’s language: Vol. 2 (pp. 127–214). New York: Gardner Press.Marchman, V. A., & Bates, E. (1994). Continuity in lexical and morphological development: A test of the critical mass hypothesis. Journal of Child Language, 21, 339–365.Marcus, G. F. (1993). Negative evidence in language acquisition. Cognition, 46, 53–85.McCartney, K. (1984). Effect of quality of day care environment on children’s language develop- ment. Developmental Psychology, 20, 244–260.McCune, L. (2006). Dynamic event words: From common cognition to varied linguistic expres- sion. First Language, 26, 233–255.Mintz, T. (2003). Frequent frames as a cue for grammatical categories in child directed speech. Cognition, 90, 91– 117.Morgan, G., Barrière, I., & Woll, B. (2006). The influence of typology and modality on the acquisi- tion of verb agreement morphology in British Sign Language. First Language, 26, 19–43.Morris, W. C., Cottrell, G. W., & Elman, J. L. (2000). A connectionist simulation of the empiri- cal acquisition of grammatical relations. In S. Wermter & R. Sun (Eds.), Hybrid neural symbolic integration (pp. 175–193). Berlin: Springer-Verlag.Mundy, P., & Gomes, A. (1998). Individual differences in joint attention skill development in the second year. Infant Behavior and Development, 21, 469–482.Naigles, L., & Hoff- Ginsberg, E. (1995). Input to verb learning: Evidence for the plausibility of syntactic bootstrapping. Developmental Psychology, 31, 827–837.Naigles, L., & Hoff-Ginsberg, E. (1998). Why are some verbs learned before other verbs? Effects of input frequency and structure on children’s early verb use. Journal of Child Language, 25, 95–120.National Institute of Child Health and Human Development Early Child Care Research Network (2000). The relation of child care to cognitive and language development. Child Development, 71, 960– 980.Newport, E. L. (1990). Maturational constraints on language learning. Cognitive Science, 14, 11–28.Newport, E. L., Gleitman, H., & Gleitman, L. R. (1977). Mother, I’d rather do it myself: Some effects and non-effects of maternal speech style. In C. E. Snow & C. A. Ferguson (Eds.), Talking to children: Language input and acquisition (pp. 109–150). Cambridge: Cambridge University Press.Ochs, E. (1985). Variation and error: A sociolinguistic approach to language acquisition in Samoa. In D. I. Slobin (Ed.), The cross-linguistic study of language acquisition: Vol. 1 (pp. 783–838). Hillsdale, NJ: Erlbaum.Pine, J. M. (1994). The language of primary caregivers. In C. Gallaway & B. J. Richards (Eds.), Input and interaction in language acquisition (pp. 15–37). Cambridge: Cambridge University Press. 136. 126 Virginia C. Mueller Gathercole and Erika HoffPinker, S. (1994). The language instinct: How the mind creates language. New York: William Morrow.Pye, C., & Quixtan Poz, P. (1988). Precocious passives (and antipassives) in Quiché Mayan. Papers and Reports on Child Language[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Development, 27, 71–80.Radford, A. (1990). Syntactic theory and the acquisition of English syntax. Oxford: Blackwell.Radford, A. (1996). Towards a structure-building model of acquisition. In H. Clahsen (Ed.), Generative perspectives on language acquisition (pp. 43–89). Amsterdam: John Benjamins.Redington, M., & Chater, N. (1997). Probabilistic and distributional approaches to language acquisition. Trends in Cognitive Sciences, 1, 273– 281.Rice, M. (1980). Cognition to language: Categories, word meaning, and training. Baltimore, MD: University Park Press.Rieckborn, S. (2006). The development of forms and functions in the acquisition of tense and aspect in German–French bilingual children. In C. Lleó (Ed.), Interfaces in multilingualism: Acquisition and representation (pp. 61–89). Amsterdam: John Benjamins.Roeper, T., & Williams, E. (Eds.). (1987). Parameter setting. Dordrecht: D. Reidel Publishing Company.Sabbagh, M. A., & Gelman, S. A. (2000) Buzzsaws and blueprints: What children need (or don’t need) to learn about language. Journal of Child Language, 27, 715–726.Saffran, J. R. (2003). Statistical language learning: Mechanisms and constraints. Current Direc- tions in Psychological Science, 12, 110–114.Saxton, M. (1997). The contrast theory of negative input. Journal of Child Language, 24, 139– 161.Saxton, M., Backley, P., & Gallaway, C. (2005). Negative input for grammatical errors: Effects after a lag of 12 weeks. Journal of Child Language, 32, 643–672.Schieffelin, B. B. (1985). The acquisition of Kaluli. In D. I. Slobin (Ed.), The cross-linguistic study of language acquisition: Vol. 1 (pp. 525–593). Hillsdale, NJ: Erlbaum.Senghas, A., & Coppola, M. (2001). Children creating language: How Nicaraguan Sign Lan- guage acquired a spatial grammar. Psychological Science, 12, 323–328.Senghas, A., Kita, S., & Özyürek, A. (2004). Children creating core properties of language: Evidence from an emerging sign language in Nicaragua. Science, 305, 1779– 1782.Shatz, M., & Ebeling, K. (1991). Patterns of language learning-related behaviours: evidence for self-help in acquiring grammar. Journal of Child Language, 18, 295–313.Shatz, M., Hoff-Ginsberg, E., & MacIver, D. (1989). Induction and the acquisition of English auxiliaries: The effects of differentially enriched input. Journal of Child Language, 16, 121–140.Shirai, Y., & Miyata, S. (2006). Does past tense marking indicate the acquisition of the concept of temporal displacement in children’s cognitive development? First Language, 26, 45–66.Slobin, D. I. (1973). Cognitive prerequisites for the development of grammar. In C. A. Ferguson & D. I. Slobin (Eds.), Studies of child language development (pp. 175–208). New York: Holt, Reinhart and Winston.Slobin, D. I. (1997). The origins of grammaticizable notions: Beyond the individual mind. In D. I. Slobin (Ed.), The cross-linguistic study of language acquisition: Vol. 5. Expanding the contexts (pp. 265–323). Mahwah, NJ: Erlbaum.Smoczynska, M. (1985). Acquisition of Polish. In D. I. Slobin (Ed.), The cross-linguistic study of language acquisition: Vol. 1 (pp. 595–686). Hillsdale, NJ: Erlbaum.Snow, C. E., Perlmann, R., & Nathan, D. (1987). Why routines are different: Toward a multiple- factors model of the relation between input and language acquisition. In K. Nelson & A. van Kleeck (Eds.), Children’s[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... language: Vol. 6 (pp. 65–98). Hillsdale, NJ: Erlbaum. 137. Input and the Acquisition of Language 127Snow, C. E., & Ferguson, C. A. (Eds.). (1977). Talking to children: Language input and acquisi- tion. Cambridge: Cambridge University Press.Soja, N. N. (1992). Inferences about the meanings of nouns: The relationship between perception and syntax. Cognitive Development, 7, 29–45.Soja, N. N., Carey, S., & Spelke, E. S. (1991). Ontological categories guide young children’s inductions of word meaning: Object terms and substances terms. Cognition, 38, 179–211.Tamis- LeMonda, C. S., Bornstein, M. H., & Baumwell, L. (2001). Maternal responsiveness and children’s achievement of language milestones. Child Development, 72, 748–767.Tavakolian, S. (1981). The conjoined clause analysis of relative clauses. In S. Tavakolian (Ed.), Language acquisition and linguistic theory (pp. 167–187). Cambridge, MA: MIT Press.Theakston, A. L., Lieven, E. V. M., Pine, J. M., & Rowland, C. F. (2002). Going, going, gone: the acquisition of the verb ‘Go’. Journal of Child Language, 29, 783– 811.Thomas, E. (2001). Aspects of gender mutation in Welsh. Unpublished PhD thesis, University of Wales Bangor.Tomasello, M. (1995). Language is not an instinct. Review of Pinker’s (1994) The language instinct: How the mind creates language (New York: William Morrow). Cognitive Development, 10, 131–156.Tomasello, M. (2000). Do young children have adult syntactic competence? Cognition, 74, 209–253.Tomasello, M. (2003). Constructing a language: A usage-based theory of language acquisition. Cambridge, MA: Harvard University Press.Waxman, S. R., Senghas, A., & Benveniste, S. (1997). A cross-linguistic examination of the noun-category bias: Its existence and specificity in French- and Spanish-speaking preschool- aged children. Cognitive Psychology, 32, 183–218.Wexler, K. (1994). Optional infinitives, head movement and the economy of derivation in child grammar. In N. Hornstein & D. Lightfoot (Eds.), Verb movement (pp. 305–350). Cambridge: Cambridge University Press.Wexler, K. (1998). Very early parameter setting and the unique checking constraint: a new explanation of the optional infinitive stage. Lingua, 106, 23–79.Wexler, K., & Manzini, M. R. (1987). Parameters and learnability in binding theory. In T. Roeper & E. Williams (Eds.), Parameter setting (pp. 41–76). Dordrecht: D. Reidel Publish- ing Company. 138. 7The Emergence of Language: A DynamicalSystems AccountJulia L. EvansThe process by which children’s language abilities emerge and the beliefs of a researchcommunity regarding the nature of language and its representation in the brain oftenare conflated in child language research. Since the late 1950s the dominant metaphorfor language and cognition has been the digital computer and the belief that humanintelligence is a process of computations on symbolic representations – rule-based manip-ulation of symbols. Language, from this perspective, is a symbolic system that is innate,residing in the human genetic code. As a result, the focus of child language research hasbeen on the universal, stable, orderly, stage-like[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... patterns in children’s language and thediscovery of these innate abstract linguistic structures. The emphasis of child language researchers has shifted recently to highlighting theflexible, transient, dynamic aspects of the emergence of language abilities. In thisendeavor, researchers are extending principles of dynamic systems theory (DST) and theself-organization of complex systems to study this process. From this perspective, lan-guage is no longer a static, abstract, symbolic system, but language patterns that emergeover time as a property of the self-organization of a complex system. Language develop-ment is no longer seen as a process of acquiring abstract rules, but as the emergence oflanguage abilities in real time, where changes over days, months, and years and moment-to-moment changes in language “processing” are the same phenomena, differing onlyin their timescales. Extending principles of DST and its emphasis on the fluid, transient,contextually sensitive nature of behavior, the goal of this approach is to identify themechanisms and states of the child’s emerging language abilities that engender develop-mental change at all levels of real-time continuous processing. The purpose of this chapter is to introduce the principles of DST and their applica-tion to language development. To set the stage, a brief overview of current theories oflanguage development is provided, along with a brief discussion of aspects of languagedevelopment not accounted for by these theories. The fundamental concepts of DST are 139. The Emergence of Language 129introduced and different modeling approaches to the study of language developmentwithin DST are presented.Historical OverviewSeveral viable accounts of language development currently exist. These accounts allendeavor to explain language development, yet they differ along several dimensionsincluding the emphasis placed on the structural versus functional aspects of children’slanguage development and on children’s competence (their knowledge of language)versus performance (actual language use). These accounts also differ with respect to thedegree to which posited mechanisms of language development are believed to residesolely in the child, the environment, or some combination of the two. Behaviorist accounts were introduced in the late 1950s, and highlight the observableand measurable aspects of language behavior. Language is viewed as a behavior like anyother behavior. Behaviorists eschew any account of language development that relies onimplicit knowledge or language competence (Mowrer, 1960; Osgood, 1963; Skinner,1957; Staats, 1971). Social-interaction accounts argue that the functional use of languagewithin the social-communicative contexts drives language development. Social-interac-tionist accounts argue that the unique form of talk directed to children by caregivers(i.e., child directed speech) is a critical part of the developmental process, with parentsproviding a tailored learning environment for the child (e.g., Bates, Beeghly-Smith,Bretherton, & McNew, 1983; Bohannon & Warren-Leubecker, 1988; Gleason, 1977;Snow, 1979, 1989). Cognitive theories acknowledge the influence of social- communicativeinteractions, but include an additional caveat – that the child[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... also brings innate cognitivecategories such as agents, patients, actions, and locations to the language learning pro-cesses (Bates, Begnigni, Bretherton, Camaioni, & Volterra, 1979; Bates & Snyder, 1987).These domain-general innate forms are semantic in nature and aid the child in interpret-ing her environment (Bowerman, 1982). Only later do abstract grammatical classes suchas noun and verb phrases emerge because of the reorganization of the innate semanticcategories. Nativist accounts view language as a species- specific, innate ability, where thechild is endowed with the grammatical structures of language, not semantic categories,which are domain-specific, genetically predetermined forms (e.g., Chomsky, 1982,1988). In the nativist accounts, emphasis is placed on identifying commonalities of lin-guistic forms both across children and across cultures as evidence of linguistic universals(Pinker, 1984).Cognitivism: The Mind-as-Computer MetaphorThe social-interactionist, cognitive, and nativist theories are strongly influenced by theassumptions of cognitive science introduced in the 1950s known as cognitivism (seeVerela, Thompson, & Rosch, 1996). Cognitivism views human intelligence as being so 140. 130 Julia L. Evanssimilar to digital computation in its essence that intelligence should be viewed as com-putations on symbolic representations. Cognitive behavior from this perspective is viewedas actions on representations that are physically realized in the form of a symbolic codein the brain or a machine, and where representations are viewed as static structures ofdiscrete symbols. Information processing from this perspective is symbolic computations – the rule-based manipulation of symbols – where the system interacts only with the form of thesymbols, not their meaning. Language, from the cognitivist perspective, is comprised ofdiscrete symbols, where it is the computations performed on the symbols that representthe meanings they stand for (see Chomsky, 1982, 1988). This abstract language knowl-edge is a symbolic code that is believed to be species-specific, and determined by humanevolution. The mind-as-computer metaphor comes with its own problems. The first is that adistinction is made between the disembodied computational mechanisms of “pure cogni-tion” and the physical implementation of behavior by the brain. This distinction resultsin a dualism between competence – the abstract, stable, enduring mental symbols thatare modular and time-independent – and performance – the “vagaries” of real- timelanguage use. In the same way that the full range of a computer program’s power is notmanifested in individual operations, it is believed that the child’s language competenceis not always displayed in real-time processing. Extracting these “essential” rules fromthe messy language performance in children has been problematic for researchers, sincethese “core abilities” seem sensitive to even the smallest changes in experimental contexts.Methodologically, this requires the construction of experimental tasks or observationalcontexts that reveal a child’s core language competence at the earliest ages possible,through the careful control of performance demands. The mind-as-computer metaphor is an excellent means to study[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... the stable, universalaspects of development, but it does not provide a framework to address certain key ques-tions. These include: Where does competence reside? If competence is genetic as Chomskysuggests (1988), how is it specified in the brain? Or as Oyama (1985) cogently argues,starting with the assumption that the code resides either in the genes or in the environ-ment inevitably leads to a logic trap – who or what turns on the genes or decides whatinformation stays and what goes? Experimentally, why should small changes in experi-mental methodologies (e.g., preferential looking versus enactment paradigms) result invastly different outcomes? What are the mechanisms that account for accurate perfor-mance under some circumstances and incorrect performance under others? Shouldn’tthese innate competences be more robust? Finally, if infants appear to have a particularlanguage skill, why is it that often this same language knowledge appears lacking inolder toddlers? The cognitivist approach also does not provide a framework for child languageresearchers to study language acquisition as a process of continuous change over time.It specifies discrete steps in the sequence of acquisition of different language skills atdifferent stages in development that occur in arbitrary timesteps. The problem is, whilelanguage development often looks like a discrete stage-like progression toward morecomplex language forms, it is often characterized by times where progress stops and even 141. The Emergence of Language 131reverts to earlier stages in development. This shifting, nonincremental nature of languagedevelopment gives rise to questions such as how does one predict a trajectory of languagedevelopment if change over time is left out of the picture? How do the moment-to-moment aspects of language learning map onto the longer timescales of development(Elman, 2003)? Where do novel language behaviors come from? How does the childmove from one stage to the next in language development? Finally, with the mind-as-computer metaphor, how are abstract symbols connectedto real-word meanings? (the symbol- grounding problem). In defining symbol representa-tions independent of their physical realizations it is not possible to know the meaningsof abstract symbols that are themselves manipulated solely on the basis of their (arbitrary)shapes, unless they are grounded in something other than more abstract symbols(Barsalou, 1999; Harnad, 1990; Searle, 1980). If the goal is to describe the underlyingmechanisms, states, and processes that engender language learning as a real-time con-tinuous process, then one needs to start at a different point conceptually. Or as Port andvan Gelder (1995, p. 2) argue, in the same way that “astronomy could only make progressby displacing the earth from the center of the universe, so must developmental sciencealso displace the inner computer from the center of cognitive performance” to under-stand language development as a continuous process. An alternative is to view language development as change over time – to assume thatthe emergence of language abilities and real-time language processing are the same phe-nomenon differing only in the timescale with which they are[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... observed. This viewassumes that language is not abstract symbols, but embodied in real life experiences(Ford & Lerner, 1992; Thelen & Smith, 1994; Verela et al., 1996). It holds that languageis softly assembled, unfolding in real time as a process of continuous, simultaneouschanges in the interactions between the environment, the body, and the nervous system– where time is always part of the equation (Elman, 2003; Port & van Gelder, 1995).Dynamic systems theory provides such a framework.Dynamic Systems TheoryDynamic systems theory (DST) is grounded in recent advances in the fields of complex,nonlinear, dynamic systems in physics and mathematics, and a longstanding traditionof systems thinking in biology and psychology. The term dynamic systems refers both tocomplex, nonlinear systems that change over time and to the formal class of mathemati-cal equations used to describe these systems. Inherent in DST are assumptions of self-organization, complexity, and emergentism. Dynamic systems theory assumes that novel,complex forms of behavior emerge from the interaction of the components of the systemand the environment. The goal in extending this approach to the study of languagedevelopment is to explain how observable language abilities evolve over time. The targetof study in this approach is the change in the behavior of complex systems, in particular,the trajectory of behaviors across different timescales. Knowledge within DST is embod-ied; it is derived from, and inextricably bound to, actions and perceptions. The defining 142. 132 Julia L. Evansproperty of development from DST is the occurrence of increasingly complex novelforms of behavior. Three assumptions characterize a DST approach to the study of development.First, developmental outcomes can be explained through the spontaneous emergence ofmore complex forms of behavior due to the cooperation of the multiple heterogeneousparts of the system that produce coherent complex patterned behavior. This process isknown as self-organization. It occurs without pre-specification from internal rules orgenetic code. Rather, development is truly self-organizing because it occurs through therecursive interactions of the components of the system. This process depends both onthe organism itself and on the constraints put on the organism by the environment inwhich it resides. This self-organization results in the emergence of novel, complex formsof behavior. Thus, developmental scientists, working within DST, see language develop-ment as a process of emergence, as opposed to growth, learning, or construction (seeLewis, 2000). Second, a DST approach assumes that self-organization is continuous intime – a dynamic process. Development is the simultaneous continuity across multiplelevels and multiple timescales. Key to this alternative view is the idea that developmentalprocesses are both nested as well as coupled across different timescales. For detaileddiscussion of DST of development, see work by Thelen, Smith, and colleagues(Corbetta & Thelen, 1996; Fogel & Thelen, 1987; Gershkoff-Stowe & Thelen, 2004;Kelso, Ding, & Schöner, 1986a; Muchisky, Gershkoff-Stowe, Cole, & Thelen, 1996;Smith & Thelen, 1993;[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Thelen, 1989; Thelen, Corbetta, & Spencer, 1996; Thelen &Fischer, 1982; Thelen, Schöner, Scheier, & Smith, 2001; Thelen & Smith, 1994; Thelen& Ulrich, 1991). Stability and instability are central to how behavioral patterns are conceptualizedwithin DST. Stability is the “persistence of behavioral or neural states in the face of sys-tematic or random perturbations” to the system (Spencer & Schöner, 2003, p. 394). Wesee examples throughout all of child language development, where children come toacquire stable language forms in the face of extreme variability in the input. For example,children are able to learn the phonemic contrasts of their language in the context ofextreme within and across speaker variability. While achieving behavioral stability iscritical in development, so is the need for flexibility and dissolution of old forms. Withthe emergence of novel, more complex forms, stable patterns must become unstable forchange to occur. This instability itself allows the components of the system to reorganizein novel ways. From a DST perspective variability is not simply “noise” in the systembut instead provides valuable insights into the nature of language development and mayin fact be the actual mechanism of change in development (Gershkoff-Stowe & Thelen,2004, p. 13). In DST the emergence of behaviors is synonymous with developmental change. Thisnotion of emergence is in fundamental contrast to nativist accounts, in that it refers tothe coming into existence of new forms or properties through the ongoing processesintrinsic to the system itself. Emergence is not a metaphor but is a general principleunderlying self- organization. Moreover, the application of the concept of emergence ofself- organization to development may inherently provide a vehicle for the unificationof ideas and thinking about the process of development across multiple domains(Lewis, 2000). 143. The Emergence of Language 133Dynamic Models of Language DevelopmentDynamic systems theory models of language and cognitive development derive fromnonequilibrium thermodynamics (Nicolis & Priogogine, 1989), catastrophe theory(Thom, 1975), synergetics (Haken, 1983, 1993, 1996), chaos theory (Ott, 1993),Hebbian learning (Hebb, 1949), and entrainment of complex systems (Kelso, 1984;Kelso et al., 1986a). These models include van der Maas and Molenaar’s use of catas-trophe theory to model cognitive development (van der Maas, 1998; van der Maas &Molenaar, 1992), van Geert’s use of ecological growth models to model language devel- opment (van Geert, 1991, 1993, 1994, 1998), connectionist models (e.g., Bates & Elman,2000; Elman, 2001, 2003; Elman et al., 1996; Rumelhart, McClelland, & Group,1986), and Thelen and colleagues’ use of Haken’s principles of synergetics (Haken, 1983,1993, 1996) and Waddington’s epigenetic landscape (Waddington, 1954, 1957, 1977) todescribe the spontaneous emergence of novel motor, cognitive, and language skills aswell as to conceptualize real-time language processing and the longer timescale of lan-guage development as integrated phenomena. Finally, both Fogel and Buder and theircolleagues have used coupled logistics equations and principles of entrainment of complexsystems to model speaker behaviors in[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... mother–child dyadic interactions (Buder, 1986,1991, 1996; Buder & Eriksson, 1997; Fogel & Thelen, 1987; Hsu & Fogel, 2003). Eachof these models is discussed below.Catastrophe modelsCatastrophe models of stage-wise cognitive development employ key concepts suchas attractor states, stable attractors, and equilibrium (e.g., van der Maas, 1998; vander Maas & Molenaar, 1992). Van der Maas, Molenaar, and colleagues argue that prin- ciples of self-organization are the only solution to answer the question of the origins ofnovel forms. In their work, they show how discontinuous jumps in development can beformally predicted by catastrophe theory. In these models, “cognitive strategies” aredefined as coherent behavioral attractors – preferred organizational states of the system.These attractors each have their own inherent stability, and can, with perturbations tothe system, be brought so far out of equilibrium that the system falls apart or “trans-forms” into a different behavioral strategy in a discontinuous manner. Reaction time isa measure of equilibrium in these models, with shorter reaction times reflecting greaterstability of the system. In catastrophe models, children will shift suddenly from one developmental state toanother when two different cognitive strategies are highly active and competing. Thiscompetition between two behavioral attractors is what causes disequilibrium of thesystem, manifesting as behavioral instability, forcing the child to reorganize into a novel,more stable state space. The onset of developmental transitions is signaled by “catastropheflags”. These “flags” are characterized by a sudden increase in behavioral variability,bimodality (e.g., shifting between two behavioral states), and slower recovery from per-turbations to the system (e.g., slower reaction times). 144. 134 Julia L. Evans Van der Maas and Molenaar focus primarily on the shifts in cognitive development;however, their work extends to language development. In their models, syntax is nolonger an abstract structure but is instead coherent behavioral attractors (Elman, 1995;Tabor, Juliano, & Tanenhaus, 1997). Accordingly, the emergence of new developmentallanguage abilities is marked by a sudden increase in behavioral variability, such as thesimultaneous use of multiple language forms from different stages in development (e.g.,“I runded, I randed, I ran from the doggie!”), and slower reaction times for processingof novel information, which is precisely what we see in children’s language development(Gershkoff-Stowe & Thelen, 2004).Logistic growth modelsVan Geert’s (1991, 1994) pioneering work adapting growth models and ecologicalsystems to model language development employs different aspects of DST. In this work,language growth is modeled using the same logistic equations that model competingresources in ecosystems. Language, in these models, is comparable to an ecosystemcomprised of various subsystems that are in a complex relationship with each other, eachvying for the finite resources that it requires to grow and change. Importantly, the car-rying capacity of the system varies with changes in the demands made by each of thesubsystems. These models show the competition and trade-offs between the demands of differentlanguage forms,[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... such as lexical and syntactic growth in children’s language development.Van Geert’s models also demonstrate tight interrelationships between phonological,lexical, and syntactic growth, as well as the continuous competition between lexical andsyntactic growth, and/or phonological and lexical growth (van Geert, 1991, 1993, 1994).These models show how logistic growth models exhibit the same stage-like behavioraltransitions in development we see in children’s emerging language abilities. Finally, vanGeert’s models show “bootstrapping” effects between phonological, lexical, and syntacticsubsystems, and “competition” for cognitive resources between subsystems, all in theabsence of abstract rules.Connectionist modelsThe focus of connectionist models is on the emergence of complex behavior in real time(Elman et al., 1996; O’Reilly & Munakata, 2000; Rumelhart & McClelland, 1986;Rumelhart et al., 1986). Using principles of neuroscience and computer simulations,connectionist accounts of language development demonstrate that seemingly rule-governed language forms emerge spontaneously from locally distributed information(e.g., Elman, 1995). From the connectionist perspective, what the child brings to theprocess of language development is neurological in nature (e.g., neurons with differentstructures found in different parts of the brain with differing firing thresholds andrefractory periods), and this constrains the child’s processing of information in the lan-guage learning environment. 145. The Emergence of Language 135 The traditional, innately determined symbolic grammars are instead represented in adistributed manner through the local connections that vary in degree of connectionweights – the micro- circuitry of the brain. Connectionist models show that the seeminglydistinct “modular” aspects of language instead simply emerge from the intrinsic proper-ties of language input. For example, children’s over-regularization of regular past tenseinflections as seen in their production of goed instead of went has been seen as evidenceof a rule-based system. However, connectionist models show that these grammaticalcontrasts can arise from the same neural network’s sensitivity to both the systematicityand frequency with which regular and irregular past tense forms occur in the extantlanguage environment (e.g., Plunkett & Marchman, 1993).SynergeticsThe work of Thelen and colleagues has its roots in synergetics and its principles of self- organization of complex systems (Haken, 1983). Synergetics focuses on the most strikingfeature of biological systems – the emergence of behavior at the macroscopic level, behav-ior absent in the individual components of the system (Haken, 1993; Nicolis &Priogogine, 1989). This is known as pattern complexity, where the collective impact ofthe aggregation of interconnected components of the system is greater than the sum ofthe individual parts. This relationship between the components of a system is viewed asa synergetic one where they are temporally assembled in a functional, task-specificmanner when under the influence of some external variable (Haken, 1996; Kelso, 1984;Kelso et al., 1986a; Kelso, Scholz, & Schöner, 1986b; Sadovsky, 1983). Since principles of complexity hold across all levels of complexity; the[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... “heuristic”value of the approach taken by Thelen and colleagues is such that the same principlesof complexity can be extended to discussions of all conventional “stages” in children’slanguage development. Development, from DST, is the emergence of “softly assembled”patterns due to the interaction between the intrinsic dynamics of the child and externalconditions. To date, Thelen, Smith and colleagues have investigated two phenomena inlanguage development – the sudden nonlinear change in the rate of children’s develop- mental trajectories of the acquisition of object labels, and U-shaped patterns in languageskills where children initially demonstrate mastery of a given language form andthen suddenly seem to lose this mastery at points in development (e.g., Plunkett &Marchman, 1991). Children produce their first words around their first birthday. Initially word learningis slow, with children requiring multiple exposures to learn the names of objects andoften needing many exemplars of a category of words before they are able to extend thename of the object to other like kinds. As children become more efficient word learners,however, the rate at which they are able to learn new words increases substantially to thepoint that they need only hear the name of a novel object once to learn its label (Mervis& Bertrand, 1994). Children also begin to correctly extend the label to other objectssharing the same physical features at this same time (Jones, Smith, & Landau, 1991;Landau, Smith, & Jones, 1988; Markman, 1989; Smith, Jones, Landau, & Gershkoff-Stowe, 2002; Waxman & Hall, 1993). 146. 136 Julia L. Evans What is curious is that when children have acquired about 75 words, their learning ofnouns based upon their shape increases their attention to shape properties of other objects,which subsequently leads to an increase in learning names for new objects having the sameshape and so on. This sudden attention to shapes of objects results in the acceleration ofchildren’s learning of object names based upon their shape (Smith, 2000). One questionis whether this attention to shape is an innate tendency, a “shape bias” at this point inlanguage development, or whether some other phenomenon is contributing to this sudden“snowball” effect. Research suggests that while this occurs naturally, children’s priorhistory of word learning can be manipulated experimentally to examine whether the“language learning history” children bring to the word-learning task affects their learningof new words (Samuelson, 2002; Samuelson & Smith, 1999, 2000; Smith, 2000; Smithet al., 2002). In a series of studies, Smith and her colleagues showed that by trainingchildren to attend to shape (i.e., artificially creating a shape bias) before the point at whichthe shape bias typically occurs in development, the rate of children’s noun learning outsidethe laboratory was accelerated by 300% in an 8-week period. This finding indicates that,at least for word learning, children’s attention to different properties of objects may notbe an innate feature but instead the collective representation of the child’s prior languagelearning history, which influences all future aspects of language acquisition. While children’s language learning can appear stage-like when viewed globally, pro-gressing toward ever-[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... increasing complexity, when observed in detail it is marked not onlyby the sudden regression back to earlier developmental stages, but often by a suddenincrease in errors before the acquisition of novel forms. While this often occurs aschildren begin to learn regular and irregular past tense verb forms (e.g., Plunkett &Marchman, 1991), another point where this occurs is when children mistakenly use thewrong name for an object for which they already know the name. This typically occurswhen children suddenly start acquiring words at an accelerated rate. These naming errorscontinue and even increase after this vocabulary spurt (Dromi, 1987). These instances when children “revert” back to performance characteristic of younger,less mature children have been interpreted in several ways: (1) the loss or abandonmentof behavior (Bever, 1982), (2) the result of experimental methods which affect perfor-mance (Klahr, 1982), or (3) as behavioral changes, which give the appearance of U-shaped growth, but which are distinct from true representational changes (Karmiloff-Smith,1992). Gershkoff- Stowe and Thelen argue that U-shaped patterns in development “arespecial cases of the nonlinearity that are the inevitable product of complex systems,composed of many, often heterogeneous parts which assemble themselves in differentconfigurations depending on the status of the components, the environment, and thetask” (Gershkoff-Stowe & Thelen, 2004, p. 12). Gershkoff-Stowe and Smith (1997) investigated the naming skills of children before,during, and after the vocabulary spurt. Their analysis revealed a sudden substantialincrease followed by an equally rapid decline in naming errors when children’s word-learning rate was at its peak, when children had approximately 75 words in their produc-tive vocabulary. The types of errors children made were in using the wrong word toname a familiar object. In particular, children would call the familiar object by a wordthey had just said which indicates that they had difficulty retrieving words from lexicalmemory. Gershkoff-Stowe and Smith suggest that these naming errors were the result 147. The Emergence of Language 137of the weakening of access of familiar words in children’s lexicons because of the rapidexpansion of the lexicon. In a second study Gershkoff-Stowe (2001) intentionally strengthened children’s rep-resentations of specific words through practice to determine if this made them less vul-nerable to interference from newly acquired words. From a DST perspective, therepresentations for newly emerging object labels are too weak to prevent the child frombeing “sucked” into older, more stable lexical states (Gershkoff-Stowe, 2001; Gershkoff-Stowe & Smith, 1997). This work suggests that word frequency may be one of the criticalfactors in accounting for naming errors at this point in development, with low-frequency,newly acquired words being more vulnerable to processing demands. Moreover, Gersh-koff-Stowe’s (2001) work shows that the child’s prior history of use of certain words (e.g.,more practiced) reduces their vulnerability to interference from other words as comparedwith less practiced words. These studies show how the types of words children learn in the first stages of devel-opment[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... create an attractor for those types of words (e.g., shape-based categories of words)that subsequently narrows their attention, shifting it away from learning other types ofwords. The emphasis of Thelen, Smith, and colleagues’ approach is on how prior learningstates influence future language learning – the developmental language trajectory. Thiswork demonstrates that at any point in time, children’s emerging language is the instan-taneous, functional, organization of the components of the child’s system to meet thedemands of real-time communication. This emergence of language is a reflection notonly of the intrinsic dynamics of the child, but also of the child’s entire language learn-ing history, current preferred states, and the extrinsic dynamics of the immediate context(Corbetta & Thelen, 1996; Fischer, Rotenberg, Bullock, & Raya, 1993; Smith & Thelen,1993; Thelen & Ulrich, 1991).An epigenetic landscape: From babbling to first wordsThelen and colleagues often use Waddington’s epigenetic landscape to conceptualizemoment-to-moment language processing and language development as integrated phe-nomena (Waddington, 1954, 1957, 1977). Figure 7.1 shows such a landscape. In thelandscape, three dimensions are represented: (1) time, (2) emergent behavior, and (3) therelative stability of the system at any point in time (e.g., depth of the attractors – preferredorganizational states of a system). Starting at the back and working forward, the surfaceof the landscape represents the irreversible process of development – the evolving attractorlandscape. The cross-section is the likelihood of an attractor state occurring at a givenmoment in time. Each of the cross-lines in the figure represents the multi-attractor char-acter of development at that point in time. The depth of the attractor landscape visuallyrepresents the stability of the system when in that self-organized state. The height of thesurface is the sensitivity of the system when in that self-organized state to external per-turbations from the environment. All of the levels are completely co-dependent. As aresult, the landscape is self-organizing through the entire developmental process. “Learning” in biological complex systems is synonymous with changes in the underly-ing attractor landscape due to input to the system (Haken, 1996). The accumulated 148. 138 Julia L. Evans Collective variable Collective variable lines Time Stability (Depth) UnderscapeFigure 7.1 Dynamic attractor landscape (Muchisky et al., 1996).effect of repeated real-time ordered states gives rise to the emergence of new collectivebehavioral states and the disappearance of developmentally older states (Haken, 1996).These ordered states will be more or less stable and coherent, and will always be a reflec-tion of the underlying architecture of an individual’s representational multi-attractorlandscape (van der Maas, 1998). Muchisky et al. (1996) use this landscape to characterize the transition from babblingto first words. In this approach, the changing landscape for speech development reflectsthe interplay between the infant’s changing articulatory system (intrinsic dynamics), theimpact of continuous language input to the infant on the infant’s underlying speechattractor landscape, and moment-to-moment extrinsic dynamics of the[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... surroundingcontext (Figure 7.2). The infant’s attractor landscape is initially comprised of three attractors: nonreflexivesounds, crying sounds, and the reflexive sounds that the infant can produce at birth. 149. The Emergence of Language 139 Nonreflexive sounds Reflexive sounds Crying sounds Phonation stage Gooing stage Expansion stage Marginal babbling Canonical babbling stage Variegated babbling Word formationFigure 7.2 Evolving speech attractor landscape (Muchisky et al., 1996).The infant’s prior state includes all the exposure to speech up to that point in time, aswell as all prior instances of sound production by the infant, all of which are continu-ously altering the properties of the underlying speech landscape. Changes in externalconditions shift the infant into and out of reflexive, nonreflexive, and crying states. Asthe infant’s vocal tract changes, vocal activity expands to include cooing and gooing,and variations in pitch and amplitude of vocalizations. New sounds appear on the speechlandscape as multi-dimensional attractors, with rhythmic stereotypic sequences occur-ring at transitions to more coordinated activity at each stage. The infants’ increasinglydetailed phonological attractor landscape eventually begins to mirror the frequency ofoccurrence of speech sounds in the language environment. Distinct phonological representational states gradually emerge in the attractor land-scape due to the accumulating vocalizing experience, resulting in the emergence of moredetailed and skilled speech production. The underlying “language” attractor landscapemirrors the properties of the language learning environment. Dynamic systems theorysimultaneously predicts global similarities in the pattern of development across children, 150. 140 Julia L. Evansmirroring the global consistency of a given language, as well as individual differencesacross children reflecting the unique, idiosyncratic, language learning environment ofeach child.Dyadic interactions and coupled dynamic complex systemsLanguage is social behavior that occurs within a communicative context (Searle, 1980).During dyadic interactions, individuals continuously alter their verbal behavior toconverge to that of their speaking partners during ongoing discourse (Beebe, Jaffe,Feldstein, Mays, & Alson, 1985; Capella, 1988; Street, 1981). One critical dimensionalong which speakers converge is timing parameters such as response duration, andspeaking rates (Capella, 1988; Street, 1981). Speakers’ attempts to converge often resultin individuals shifting from their preferred speaking rates to a third rate, that of thedyad. Developmental differences exist in the degree to which children shift to the adult’srate, with children as young as 5 attempting to match an adult’s turn durations, responselatencies, internal pauses, and speech rates (Welkowitz, Cariffe, & Feldstein, 1976). In nature, complex systems constantly interact with each other. When the elementsof one complex system influence the state of another system, we can talk about couplingof complex systems. Two systems are “coupled” if changes in the value of a parameterin one system influence the value of the same parameter in a second system. This isknown as[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... entrainment. Nonlinear dynamic models of the dyadic interactions of twoindependent complex systems show that (1) biological systems have a strong tendencyto converge or coordinate at many different behavioral levels (Kelso et al., 1986a), and(2) changes in the behavior of one system can result in instability in the behavior of thesecond system (Buder, 1991). Research on the coupling or entrainment of two systemsindicates that the problem of convergence is solved adaptively, whereby two systems altertemporal features to reach mutual relative coordination. Thus, entrained systems shouldbe expected to match only along some dimensions, but not all dimensions. Buder’s (1991) logistic model of dyadic discourse interactions shows how the impact ofchanges in one speaker’s behavior on another changes over time. The model captures howspeakers converge on a range of discourse parameters such as degree of desire to participateand degree of other relatedness. When both speakers in the dyad have the same desire toparticipate in the interaction, the model converges easily. However, the model also showsthat when the behavior of one speaker is altered continuously along a dimension (e.g., lackof willingness to participate, or no response on the listener’s part) to the point where itreaches a critical threshold, the behavior of the second speaker becomes erratic and neitherspeaker in the dyad is able to converge. The instability and chaotic behavior of the dyadgenerated by the Buder (1991) model has been observed in the phase portraits of turn-by-turn behavior of children with specific language impairments who fail to meet the timingdemands of dyadic interactions with an adult speaking partner. The result is the continu-ous breakdown in turn-taking between the adult examiner and the child when in conversa-tion with children with language disorders (Evans, 1998, 2002). These techniques, which capture the behavior of coupled complex systems, alsoprovide the means to mathematically capture the state of the individual child’s verbal 151. The Emergence of Language 141system as it is shifted out of its preferred processing “state space” due to changes in the“external” discourse demands. They also show the degree to which caregivers and infantscan converge along verbal and nonverbal dimensions during dyadic interactions (e.g.,Fogel & Thelen, 1987; Hsu & Fogel, 2003; Newtson, 1998). These studies have impor-tant methodological implications. One common method of studying language develop-ment is to record children’s spontaneous language during examiner–child, or parent–child,dyads and then study only the child’s language productions in isolation. However, non-linear dynamic models clearly show that there is no point at which children’s languagecan be studied independent of the context within which the data are collected. Each of the lines of work discussed so far demonstrate how, at any point in develop- ment, the emergence of language is the instantaneous, functional, organization of thecomponents of the child’s language system to meet the real-time communicative demands.Language abilities are a reflection not only of the intrinsic dynamics of the child, butalso of the child’s prior language[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... learning history, current preferred states, and theextrinsic dynamics of the immediate context.Mental Representations: Dynamic Field Theoryand ConnectionismAny study of language development has to address the mind’s representation of language.Earlier work in DST assumed that representations, if they even existed, were unlikelyto play the kind of role in the brain that they play in the structure of computationalsystems (see Port & van Gelder, 1995). In contrast to connectionism, earlier DSTapproaches did not have a formalized theory of mental representations and learning.Dynamic field theory (DFT) addresses the issue of representational states (Erlhagen,Bastian, Jancke, Riehle, & Schöner, 1999; Erlhagen & Schöner, 2002; Schöner, Kopecz,& Erlhagen, 1997; Schutte & Spencer, 2002; Schutte, Spencer, & Schöner, 2003; Thelenet al., 2001). In DFT, stability is a central tenet, allowing representational states to emerge fromsensory-motor origins. Dynamic field theory makes a critical distinction between repre-sentations and representational states, however, invoking the time-dependent concept ofactivation consistent with mathematical psychology, connectionism, and theoreticalneuroscience (Bates & Elman, 2000; Churchland & Sejnowski, 1992; Williams, 1986).Dynamic field theory grounds the representational states in the pre-cognitive era of the1940s and 1950s where representational states were seen as actually re-presenting eventsof the environment in the nervous system. In DFT, representational states are activationfields that are mathematically defined across the dimensions being represented. To date,DFT has been extended primarily to the study of infant and children’s spatial workingmemory for location (Schutte & Spencer, 2002; Schutte et al., 2003) and infant’s per-severative reaching (e.g., Thelen et al., 2001). However, the use of a model that reliesboth on distributed representations while simultaneously incorporating changes in rep-resentational states via working memory stands to integrate connectionist modeling andclassical DST into a more complete DST account of language development. 152. 142 Julia L. EvansConclusionThe predominant cognitivist approaches provide little guidance to the study of languagecompetence and performance as an integrate phenomenon which emerges in real timeas a continual process of change. In DST, moment-to-moment language processing andthe longer timecourse of development are the same phenomenon simply viewed on dif-ferent timescales. Real-time language processing is the spontaneous self-organization ofthe system as it moves into and out of different meaning attractors, whereas languagedevelopment is the change in this same underlying attractor language landscape due toinput to the system. There is no distinction between “competence” and “performance”in DST. Instead, language is always and only performance within context. As Thelen(1995) notes, “even though some preferred states of the system will be so stable that theymay ‘look’ like they are the result of symbolic rules, or stages in development, the stabilityof the child’s multi-attractor language landscape is a function of the child-in-context,always. In other words, development looks stage-like only because in the immediateassembly of the activity within[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... context, certain patterns are strongly preferred” (p. 77). Dynamic systems theory is the study of self-organized, emerging patterns of complexsystems. Its strength is its focus on the stability and the variability of emerging languageabilities. Dynamic systems theory provides a set of assumptions regarding how languageis organized and changes over time. It provides an approach to empirically unlock thenature of these patterns as they change in real time as well as a framework to integratethe study of development from the neurophysiology of embodied representations asintentional communication grounded within the socially communicative context. It doesnot tell researchers which language patterns to study, the nature of the interactionbetween the child and the observational context, nor the key features of the child’s priorlearning history that influence later language learning. These must be discovered.Dynamic systems theory does provide a clear starting point, however. It is at points ofinstability – the noisy, messy, unpredictable points in the emergence of language – thatthe underlying dynamics of the emergence of language in children will be revealed. Bynot holding experimental conditions constant, but instead intentionally changing experi-mental conditions in clearly specified ways, researchers will be able to understand morefully the process of the emergence of language in children.ReferencesBarsalou, L. W. (1999). Perceptual symbol systems. Behavioral and Brain Sciences, 22, 577–660.Bates, E., Beeghly-Smith, L., Bretherton, I., & McNew, S. (1983). Social basis of language development: A reassessment. In H. Reese & L. P. Lipsitt (Eds.), Advances in child development and behavior: Vol. 16 (pp. 8–75). New York: Academic Press.Bates, E., Begnigni, L., Bretherton, I., Camaioni, L., & Volterra, V. (1979). The emergence of symbols: Cognitive and communication in infancy. New York: Academic Press.Bates, E., & Elman, J. (2000). The ontogeny and phylogeny of language: A neural network per- spective. In S. T. Parker & J. Langer (Eds.), Biology, brains, and behavior: The evolution of human development (pp. 89–130). Santa Fe, NM: School of American Research Press. 153. The Emergence of Language 143Bates, E., & Snyder, L. (1987). The cognitive hypothesis in language development. In I. Uzgiris & J. McV. Hunt (Eds.), Research with scales of psychological development in infancy (pp. 168– 206). Champaign, IL: University of Illinois Press.Beebe, B., Jaffe, J., Feldstein, S., Mays, K., & Alson, D. (1985). Interpersonal timing: The applications of an adult dialogue model to mother–infant vocal and kinesthetic interac- tions. In T. Field & N. Fox (Eds.), Social perception in infants (pp. 217–247). Norwood, NJ: Ablex.Bever, T. G. (1982). Regressions in mental development: Basic phenomena and theories. Hillsdale, NJ: Lawrence Erlbaum Associates.Bohannon, J. N., & Warren-Leubecker, A. (1988). Recent development in child-directed speech: You’ve come a long way, baby-talk. Language Sciences, 10, 89–110.Bowerman, M. (1982). Re-organizational processes in lexical and syntactic development. In E. Wanner & I. Gleitman (Eds.), Language acquisition: The state of the art. Cambridge: Cambridge[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... University Press.Buder, E. H. (1986). Coherence of speech rhythms in conversations: Autocorrelation analysis of fundamental voice frequency. Toronto Semiotic Circle Monograph.Buder, E. H. (1991). A nonlinear dynamic model of social interaction. Communication Research, 18, 174–198.Buder, E. H. (1996). Dynamics of speech processes in dyadic interactions. In J. H. Watt & C. A. VanLear (Eds.), Dynamic patterns in communication processes (pp. 301–325). Thousand Oaks, CA: Sage.Buder, E. H., & Eriksson, A. (1997). Prosodic cycles and interpersonal synchrony in American English and Swedish. Proc. Eurospeech, 1, 235–238.Capella, J. (1988). Interaction patterns and social interpersonal relationships. In S. Duck (Ed.), Handbook of social and personal relationships. New York: Guildford Press.Chomsky, N. (1982). Lectures on government and binding. New York: Foris.Chomsky, N. (1988). Language and the problems of knowledge. Cambridge, MA: MIT Press.Churchland, P. S., & Sejnowski, T. J. (1992). The computational brain. Cambridge, MA: Bradford Book/MIT Press.Corbetta, D., & Thelen, E. (1996). The developmental origins of bimanual coordination: A dynamic systems perspective. Journal of Experimental Psychology: Human Perception and Per- formance, 22, 502–522.Dromi, E. (1987). Early lexical development. Cambridge: Cambridge University Press.Elman, J. (1995). Language as a dynamical system. In R. F. Port & T. van Gelder (Eds.), Mind as motion. Cambridge, MA: MIT Press.Elman, J. (2001). Connectionism and language acquisition. In M. Tomasello & E. Bates (Eds.), Language development: The essential readings (pp. 295–306). Malden, MA: Blackwell Publishers.Elman, J. (2003). Development: It’s about time. Developmental Science, 6, 430– 433.Elman, J., Bates, E., Johnson, M., Karmiloff-Smith, A., Parisi, D., & Plunkett, K. (1996). Rethinking innateness: A connectionist perspective on development. Cambridge, MA: MIT Press.Erlhagen, W., Bastian, A., Jancke, D., Riehle, A., & Schöner, G. (1999). The distribution of neuronal population activation (dpa) as a tool to study interactions and integration in cortical representations. Journal of Neural-science Methods, 94, 53–66.Erlhagen, W., & Schöner, G. (2002). Dynamic field theory of movement preparation. Psychologi- cal Review, 109, 545–572.Evans, J. L. (1998). Verbal and nonverbal discourse cues in nonlinear dynamic models of discourse. American Psychological Associate Annual Meeting, Washington, DC. 154. 144 Julia L. EvansEvans, J. L. (2002). Variability in comprehension strategy use in children with SLI: A dynamical systems account. International Journal of Language and Communication Disorders, 37, 95–116.Fischer, K. W., Rotenberg, E. J., Bullock, D. H., & Raya, P. (1993). The dynamics of compe- tence: How context contributes directly to skill. In R. H. Wozniak & K. W. Fischer (Eds.), Development in context: Acting and thinking in specific environments (pp. 93–117). Hillsdale, NJ: Lawrence Erlbaum Associates.Fogel, A., & Thelen, E. (1987). Development of early expressive and communicative action: Reinterpreting the evidence from a dynamic systems perspective. Developmental Psychology, 23, 747–761.Ford, D. H., & Lerner, R. M. (1992). Developmental systems theory: An integrative approach.[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Newbury Park, CA: Sage.Gershkoff-Stowe, L. (2001). The course of children’s naming errors in early word learning. Journal of Cognition and Development, 2, 131–155.Gershkoff-Stowe, L., & Smith, L. B. (1997). A curvilinear trend in naming errors as a function of early vocabulary growth. Cognitive Psychology, 34, 37–71.Gershkoff-Stowe, L., & Thelen, E. (2004). U-shaped changes in behavior: A dynamic systems approach. Journal of Cognition and Development, 5, 11–36.Gleason, J. B. (1977). Some notes on feedback. In C. E. Snow & C. A. Ferguson (Eds.), Talking to children: Language input and acquisition. Cambridge: Cambridge University Press.Haken, H. (1983). Synergetics, an introduction. New York: Springer- Verlag.Haken, H. (1993). Advance synergetics: Instability hierarchies of self- organizing systems and devices. New York: Springer-Verlag.Haken, H. (1996). Principles of brain functioning: A synergetic approach to brain activity, behavior and cognition. Berlin, Germany: Springer-Verlag.Harnad, S. (1990). The symbol grounding problem. Physica, D, 42, 335–346.Hebb, D. O. (1949). The organization of behavior. New York: Wiley.Hsu, H. C., & Fogel, A. (2003). Stability and transitions in mother–infant face-to-face com- munication during the first 6 months: A micro-historical approach. Developmental Psychology, 39, 1061–1082.Jones, S. S., Smith, L. B., & Landau, B. (1991). Object properties and knowledge in early lexical learning. Child Development, 62, 499–516.Karmiloff-Smith, A. (1992). Beyond modularity: A developmental perspective on cognitive science. Cambridge, MA: MIT Press.Kelso, J. A. S. (1984). Phase transitions and critical behavior in human bimanual coordination. American Journal of Physiology: Regulatory, Integrative, and Comparative Physiology, 15, 1000–1004.Kelso, J. A. S., Ding, M., & Schöner, G. (1986a). Dynamic pattern formation: A tutorial. In L. B. Smith & E. Thelen (Eds.), A dynamic systems approach to development: Applications (pp. 13– 50). Boston, MA: MIT Press.Kelso, J. A. S., Scholz, J. P., & Schöner, G. (1986b). Non-equilibrium phase transitions in coor- dinated biological motion: Critical fluctuations. Physic Letters A, 118, 279–284.Klahr, D. (1982). Non- monotone assessment of monotone development: An information process- ing analysis. In S. Strauss (Ed.), U-shaped behavioral growth (pp. 63–86). New York: Academic Press.Landau, B., Smith, L. B., & Jones, S. S. (1988). The importance of shape in early lexical learn- ing. Cognitive Development, 3, 299–321.Lewis, M. D. (2000). The promise of dynamic systems approaches for an integrated account of human development. Child Development, 71, 36– 43. 155. The Emergence of Language 145Markman, E. M. (1989). Categorization and naming in children. Cambridge, MA: MIT Press.Mervis, C. D., & Bertrand, J. (1994). Acquisition of the novel name–nameless category principle. Child Development, 65, 1646–1662.Mowrer, O. H. (1960). Learning theory and the symbolic process. New York: Wiley.Muchisky, M., Gershkoff-Stowe, L., Cole, E., & Thelen, E. (Eds.). (1996). The epigenetic land- scape revisited: A dynamic interpretation: Vol. 10. Greenwood Publishing.Newtson, D. (1998). Dynamical systems and the structure of behavior. In K. M. Newell & P. C.[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Molenaar (Eds.), Applications of nonlinear dynamics to developmental process modeling (pp. 199–220). Mahwah, NJ: Lawrence Erlbaum Associates.Nicolis, G., & Priogogine, I. (1989). Exploring complexity: An introduction. New York: W. H. Freeman & Company.O’Reilly, R. C., & Munakata, Y. (2000). Computational explorations in cognitive neuroscience: Understanding the mind by simulating the brain. Cambridge, MA: MIT Press.Osgood, C. (1963). Method and theory in experimental psychology. New York: Oxford University Press.Ott, E. (1993). Chaos in dynamical systems. Cambridge: Cambridge University Press.Oyama, S. (1985). The ontogeny of information: Developmental systems and evolution. Cambridge, MA: Cambridge University Press.Pinker, S. (1984). Language learnability and language development. Cambridge, MA: Harvard University Press.Plunkett, K., & Marchman, V. (1991). U-shaped learning and frequency effects in a multi-layered perceptron: Implications for child language acquisition. Cognition, 38, 43–102.Plunkett, K., & Marchman, V. (1993). From rote learning to system building: Acquiring verb morphology in children and connectionist nets. Cognition, 48, 21–69.Port, R. F., & van Gelder, T. (1995). Mind as motion. Cambridge, MA: MIT Press.Rumelhart, D. E., & McClelland, R. J. (1986). On learning the past tenses of English verbs. In R. J. McClelland & D. E. Rumelhart (Eds.), Parallel distributed processing: Explorations in the microstructure of cognition: Vol. 2. Cambridge, MA: MIT Press.Rumelhart, D. E., McClelland, R. J., & Group, T. P. R. (Eds.). (1986). Parallel distributed pro- cessing: Explorations in the microstructure of cognition: Vol. 1. Foundations. Cambridge, MA: MIT Press.Sadovsky, N. N. (1983). Systems theory. In R. Harré & R. Lamb (Eds.), The encyclopedic diction- ary of psychology (pp. 623–625). Oxford: Blackwell.Samuelson, L. K. (2002). Statistical regularities in vocabulary guide language acquisition in connectionist models and 15–20-month-olds. Developmental Psychology, 36, 1016–1037.Samuelson, L. K., & Smith, L. B. (1999). Early noun vocabularies: Do ontology, category, orga- nization and syntax correspond? Cognition, 73, 1–33.Samuelson, L. K., & Smith, L. B. (2000). Children’s attention to rigid and deformable shape in naming and non-naming tasks. Child Development, 71, 1555–1570.Schöner, G., Kopecz, K., & Erlhagen, W. (1997). The dynamic neural field theory of motor programming: Arm and eye movements. In P. G. Morasso & V. Sanguineti (Eds.), Self- organization, computational maps and motor control: Vol. 119 (pp. 271–310). Amsterdam: Elsevier-North Holland.Schutte, A. R., & Spencer, J. (2002). Generalizing the dynamic field theory of the a-not-b error beyond infancy: Three-year-olds’ delay- and experience-dependent location memory biases. Child Development, 73, 377– 404.Schutte, A. R., Spencer, J., & Schöner, G. (2003). Testing the dynamic field theory: Working memory for locations becomes more spatially precise over development. Child Development, 74, 1393–1417. 156. 146 Julia L. EvansSearle, J. R. (1980). Minds, brains and programs. Behavioral and Brain Sciences, 3, 417–457.Skinner, B. F. (1957). Verbal behavior. Englewood Cliffs, NJ: Prentice-Hall.Smith, L. B. (2000). Learning[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... how to learn words: An associative crane. In K. Hirsh-Pasek (Ed.), Becoming a word learner: A debate on lexical acquisition (pp. 51–80). New York: Oxford University Press.Smith, L. B., Jones, S. S., Landau, B., & Gershkoff- Stowe, L. (2002). Object naming learning provides on-the-job training for attention. Psychological Science, 13, 13–19.Smith, L. B., & Thelen, E. (Eds.). (1993). A dynamic systems approach to development: Applica- tions. Cambridge, MA: MIT Press.Snow, C. E. (1979). The role of social interaction in language acquisition. In W. A. Collins (Ed.), Minnesota symposia on child psychology: Vol. 12. Hillsdale, NJ: Lawrence Erlbaum Associates.Snow, C. E. (1989). Understanding social interaction and language acquisition: Sentences are not enough. In M. Bornstein & J. Brunner (Eds.), Interactions in human development (pp. 83–104). Hillsdale, NJ: Lawrence Erlbaum Associates.Spencer, J., & Schöner, G. (2003). Bridging the representational gap in the dynamic systems approach to development. Developmental Science, 6, 392–412.Staats, A. (1971). Linguistic-mentalistic theory versus an explanatory s-r learning theory of lan- guage development. In D. I. Slobin (Ed.), The ontogenesis of grammar. New York: Academic Press.Street, R. L. (1981). Evaluation of convergence in mother–child conversations. Social Communi- cation, 36, 87–95.Tabor, W., Juliano, C., & Tanenhaus, M. (1997). Parsing in dynamical systems: An attractor- based account of the interaction of lexical and structural constraints in sentence processing. Language and Cognitive Processes, 12, 211–271.Thelen, E. (1989). Self-organization in developmental processes: Can systems approaches work? In M. Gunnar & E. Thelen (Eds.), Systems and development: Minnesota symposia on child psy- chology: Vol. 22 (pp. 77–117). Hillsdale, NJ: Lawrence Erlbaum Associates.Thelen, E. (1995). Time-scale dynamics and the development of an embodied cognition. In R. F. Port & T. van Gelder (Eds.), Mind in motion. Cambridge, MA: MIT Press.Thelen, E., Corbetta, D., & Spencer, J. (1996). The development of reaching during the first year: The role of movement speed. Journal of Experimental Psychology: Human Perception and Performance, 22, 1059–1076.Thelen, E., & Fischer, D. M. (1982). Newborn stepping: An explanation for a “disappearing reflex.” Developmental Psychology, 18, 760–770.Thelen, E., Schöner, G., Scheier, C., & Smith, L. B. (2001). The dynamics of embodiment: A field theory of infant perseverative reaching. Behavioral and Brain Sciences, 24, 1–86.Thelen, E., & Smith, L. B. (1994). A dynamic systems approach to the development of cognition and action. Cambridge, MA: MIT Press.Thelen, E., & Ulrich, B. D. (1991). Hidden skills: A dynamical systems analysis of treadmill stepping during the first year. Monographs of the Society for Research in Child Development, 56 (233).Thom, R. (1975). Structural stability and morphogenesis. Reading: Benjamin.Van der Maas, H. L. J. (1998). The dynamical and statistical properties of cognitive strategies: Relations between strategies, attractors, and latent classes. In K. M. Newell & P. C. M. Molenaar (Eds.), Applications of nonlinear dynamics to developmental process modeling (pp. 161–176). Mahwah, NJ: Lawrence Erlbaum Associates.Van der Maas, H. L. J., &[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Molenaar, P. C. (1992). Stage-wise cognitive development: An applica- tion of catastrophe theory. Psychological Review, 99, 395–417. 157. The Emergence of Language 147Van Geert, P. (1991). A dynamic systems model of cognitive and language growth. Psychological Review, 98, 3–53.Van Geert, P. (1993). A dynamic systems model of cognitive growth: Competition and support under limited resources conditions. In L. B. Smith & E. Thelen (Eds.), A dynamic systems approach to development: Applications. Cambridge, MA: MIT Press.Van Geert, P. (1994). Dynamic systems of development. London: Harvester Wheatsheaf.Van Geert, P. (1998). A dynamic systems model of cognitive and language growth. Psychological Review, 98, 3–53.Verela, F. J., Thompson, E., & Rosch, E. (1996). The embodied mind. Cambridge, MA: MIT Press.Waddington, C. H. (1954). The integration of gene-controlled processes and its bearing on evolu- tion. Proceedings of the 9th International Conference of Genetics, 9, 232– 245.Waddington, C. H. (1957). The strategy of the genes. London: Allen & Unwin.Waddington, C. H. (1977). Tools for thought. London: Allen & Unwin.Waxman, S. R., & Hall, D. G. (1993). The development of a linkage between count nouns and object categories: Evidence from fifteen- to twenty- one-month-old infants. Child Development, 64, 1224–1241.Welkowitz, J., Cariffe, G., & Feldstein, S. (1976). Conversational congruence as a criterion of socialization in children. Child Development, 47, 296–272.Williams, R. J. (1986). The logic of activation functions. In D. E. Rumelhart, R. J. McClelland & T. P. R. Group (Eds.), Parallel distributed processing: Explorations in the microstructure of cognition: Vol. 1. Foundations (pp. 423–443). Cambridge, MA: MIT Press. 158. PART IILanguage Development in Infancy 159. IntroductionThe ingredients that produce language acquisition, when combined, are a human infantand exposure to speech. The chapters in this Part ask what in human infants, what inthe speech they hear, and what internal processes produce this reliably obtained result.Four chapters address these questions, describing the infant’s starting state and the pro-cesses and outcomes of the language learning that takes place in the first years of life.Polka, Rvachew, and Mattock describe the development of speech perception and pro-duction as resulting from the interaction of three factors: infants’ initial perceptualcapacities and biases, the language(s) to which infants are exposed, and infants’ percep-tions of the sounds they themselves produce. On the perception side, intake capacitiesand input properties tune an “emerging language filter” with the result that infantsbecome less sensitive to some contrasts the language does not use and more sensitive tosome the language does use. The full developmental pattern is more complex than this,however, and Polka et al. also set out for us the phenomena that, at this point, await anexplanatory framework. On the production side, Polka et al. describe a process in whichinfants use adult input to form representations of the target sounds and use their percep-tions of self-produced speech to[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... adapt their articulatory gestures to produce thosetargets. Gerken describes infants as bringing to the language learning task the capacity todetect patterns in input and to generalize from these detected patterns, given sufficientevidence. As a result of applying these capacities to the input they receive, infants knowabout the sound sequences that make possible words in their language and somethingabout the word sequences that make possible sentences in their language well before theyproduce these forms or understand their meanings. Whereas the chapters by Polka et al. and Gerken describe the infant’s acquisition oflanguage form, the chapters by Poulin- Dubois and Graham and by Naigles and Swensenaddress how infants (and toddlers) learn the mappings from language forms to meaning. 160. 152 Part II: IntroductionPoulin-Dubois and Graham review evidence regarding the cognitive understandings thatinfants bring to the task of learning words. Before acquiring language, infants organizetheir world into individuated objects and they parse the ongoing flow of activity intoseparable events. This cognitive organization supports, but does not explain, lexicaldevelopment. Poulin-Dubois and Graham also describe effects of language on categoriza-tion and developmental changes in how children bring nonlinguistic processes to bearas they acquire the vocabulary of their language. Naigles and Swensen make the argument that very young children approach the wordlearning task already knowing something about syntax and the correspondences betweensyntax and semantics. Some syntax/semantics correspondences might be universal andavailable to children before 18 months; others are language specific and must be learned,but even these are operating at 24 months. For the child who can make use of syntax/semantics correspondences, the task of figuring out the meaning of newly encounteredwords is not just a matter of finding the referent in the world. There are clues to wordmeaning in the syntactic frame that surrounds the word. Together these chapters describe a very competent infant who enters the world withperceptual and learning capacities that bias the infant to attend to speech, to extractpatterns from that to which they attend, and to generalize these patterns. As a result,very young children have the ability to segment the speech stream and to recognize therecurring sequences that are the words and sentences of their language. As they beginthe task of finding the meanings of the words they encounter, infants rely on both theircognitive organization of the world and the information provided by the structures inwhich words appear. 161. 8Experiential Influences on Speech Perceptionand Speech Production in InfancyLinda Polka, Susan Rvachew, and Karen MattockMature language users are highly specialized, expert, and efficient perceivers and produc-ers of their native language. This expertise begins to develop in infancy, a time whenthe infant acquires language-specific perception of native language phonetic categoriesand learns to produce speech-like syllables in the form of canonical babble. The emer-gence of these skills is well described by past[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... research but the precise mechanisms bywhich these foundational abilities develop have not been identified. This chapter provides an overview of what is currently known about the impact of lan-guage experience on the development of speech perception and production during infancy.Throughout we affirm that experiential influences on phonetic development cannot beunderstood without considering the interaction between the constraints that the childbrings to the task and the nature of the environmental input. In the perception and pro-duction domains our current understanding of this interaction is incomplete and tends tofocus on the child as a passive receiver of input. In our review, we signal a recent shift inresearch attention to the infant’s role in actively selecting and learning from the input. We begin this chapter by describing what is currently known about the determinantsof speech perception and speech production development during infancy while high-lighting important gaps to be filled within each domain. We close by emphasizing theneed to integrate research across the perception and production domains.Speech Perception DevelopmentSpeech perception development involves a complex interaction between the child andhis/her language environment(s). In this section we discuss when and how languageinput begins to shape speech perception and then consider how speech intake is directed 162. 154 Linda Polka et al.and constrained from the infant side of the interaction. This discussion focuses on thedevelopment of speech perception at the segmental or phonetic level, that is, the percep-tion of consonants, vowels, and tones.Effects of age and language inputDevelopmental cross- linguistic research has shown that from the first few months of lifethe ambient language begins to guide the infant toward the goal of language- specificspeech perception. Werker and collaborators conducted the first systematic investiga-tions of the effects of age and language experience on phonetic perception. In a seriesof studies, they found that 6- to 8-month-old English infants and native adult speakersdiscriminated Hindi and Salish contrasts equally well, but discrimination declinedwith age for infants tested at 8 to 10 and at 10 to 12 months and was poor for Englishadults; no decline in discrimination over age was reported for Hindi and Salish infants(Werker, Gilbert, Humphrey, & Tees, 1981; Werker & Tees, 1983, 1984a). These find- ings showed that infants initially respond to phonetic differences in a language-neutralway but their perception becomes more language-specific by the end of the first year oflife. Werker’s findings were initially interpreted as showing that the absence of languageexperience results in a “loss” of perceptual function. This meshed well with the idea thatlanguage experience sets up a perceptual filter for speech, an idea that first emerged toexplain poor perception of non-native phonetic contrasts by adults (e.g., Miyawaki etal., 1975). Subsequent developmental cross-linguistic research expanded to include awider variety of phonetic contrasts from many languages, including consonants, vowels,and tones, as well as more varied infant groups, including infants acquiring languagesother than English and infants learning more than[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... one language simultaneously. Nowthat we have a more detailed picture of the changes in phonetic perception that areshaped by experience with a specific language during infancy, it is time to reassess thefilter analogy. For this reason, we will summarize developmental and cross-linguisticresearch by considering what this work tells us about the functional properties of thenative language filter that is emerging in infancy. In the current literature it is clear thatthe process by which infants arrive at language-specific phonetic perception is morecomplex than the original interpretation of the early studies by Werker and colleagues.The diverse developmental patterns of speech perception that we describe show that theemerging filter is active and can operate to facilitate as well as to attenuate access tophonetic differences. Furthermore, these patterns reflect an emerging native languagefilter that is sensitive to the demands of the perceptual task (discrimination vs. labeling),the stimulus domain of the test materials (speech vs. non-speech), and the functionalstatus of the phonetic elements in the native language (phonemic vs. non-phonemic;segmental vs. non-segmental).Developmental patterns in non-native phonetic perception. The initial view of a “loss” ofnon- native phonetic perception was re-interpreted when it was found that adult percep-tion of native and non-native contrasts is comparable when task demands are reduced 163. Infants’ Speech Perception and Production 155(e.g., Werker & Logan, 1985; Werker & Tees, 1984b), therefore confirming that theperceptual filter established by language experience is not absolute. Similarly, infantspeech perception patterns are not so simple. Research clearly indicated that the nativelanguage filter emerging in infancy is domain-specific and does not impact how infantsperceive sounds that are not recognized as speech. The first cross-linguistic evidence tosupport this claim was provided by Best, McRoberts, and Sithole (1988). They foundthat English infants failed to show a decline in discrimination of a non-native consonantcontrast (lateral–apical clicks) from the Zulu language. The domain-specificity of thenative language filter in infants is also confirmed by a study of tone perception in infants.In this study 6- and 9-month-old non-tone (English) and tone language infants (Chinese:Cantonese or Mandarin) were tested on discrimination of lexical tone contrasts cuedmostly by fundamental frequency (F0: perceived as speakers’ pitch) and non-speech toneanalogs with the same F0 differences (Mattock & Burnham, 2006). There was a declinein discrimination over age for lexical tone contrasts in English but not Chinese infants.Discrimination level for non-speech tone was maintained over age for both languagegroups. Mattock and Burnham’s work (2006) shows that the native language filter operatesin a functionally specific way – the ability to interpret F0 as a segmental cue was main-tained only by infants acquiring a language in which F0 has a segmental (as opposed toa suprasegmental) function. Functional specificity of the native language filter is alsosupported by findings of Pegg and Werker (1997) showing that infants also fail tomaintain perception of an allophonic difference, a phonetic difference that occursregularly in the[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... ambient language but is not used to signal differences in word meaning.They found that English infants’ perception of two allophones of English /p/ – avoiceless aspirated bilabial stop (e.g., the first consonant in “pie”) and a voiceless unaspi-rated bilabial stop (e.g., the second consonant in “spy”) – declined between 6–8 and10–12 months of age despite their exposure to each phone in different phonologicalcontexts. Cross-linguistic research on the perception of segmental contrasts reveals differencesin the timing of the perceptual decline across different non-native contrasts – 6 to 12months for consonant contrasts (Best, 1995; Best et al., 1990; Best, McRoberts, LaFleur,& Silver-Isenstadt, 1995; Bosch & Sebastián-Gallés, 2003b; Eilers, Gavin, & Oller,1982; Tsao, Liu, Kuhl, & Tseng, 2000; Tsushima et al., 1994), 6 to 8 months for vowelcontrasts (Best et al., 1997; Bosch & Sebastián- Gallés, 2003a; Polka & Werker, 1994),and 6 to 9 months for lexical tone1 (Mattock & Burnham, 2006). Although attenuation with age is often observed for perception of non-native con-trasts, this is not the only reported developmental pattern. Some non-native contrastsremain discriminable through the native language filter, including certain consonant(see Best, 1991, 1995; Best et al., 1990; Polka, Colantonio, & Sundara, 2001) and vowelcontrasts (see Polka & Bohn, 1996). In these studies discrimination levels are consis-tently quite high across development.2 It is not yet clear what these findings tell us aboutthe emerging native language filter.Developmental patterns in native phonetic perception. There is increasing evidencethat varied developmental patterns are also observed for native language phonetic percep- 164. 156 Linda Polka et al.tion. We observed a facilitative effect of native language experience in our lab when wecompared monolingual English and monolingual French listeners in four age groups(6–8 months; 10–12 months; 4 years; adults) on their perception of English /d/ versus/ð/, a contrast that is not phonemic in French. English perceivers showed significantimprovement in discrimination of the English /d–ð/ contrast between 10–12 monthsand 4 years, and further improvement between 4 years and adulthood; French perceiversshowed no change with age (Polka et al., 2001; Sundara, Polka, & Genesee, 2006). Tsaoet al. (2000) also found evidence for native language facilitation in Chinese-learninginfants’ perception of a native fricative–affricate contrast. These findings challenge thelong established view that language experience serves only to prevent a developmentaldecline in perceptual discrimination of some contrasts. The evidence for facilitation aswell as maintenance demonstrates that the language filter is active: it can enhance aswell as reduce accessibility to phonetic contrasts. An age-related decline in discrimination in infancy has also been found for a nativephonetic contrast, specifically English-learning infants’ discrimination of the native /s–z/fricative voicing contrast (Best & McRoberts, 2003). This finding indicates that somephonetic differences are less salient than others irrespective of language experience,leading to different developmental patterns within the native language. Moreover, find-ings also show that infant perception of native[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... contrasts also depends on the task athand. For example, Stager and Werker (1997) observed that 14-month-old English-learn-ing infants were unable to detect a native /bI–dI/ contrast when each syllable was pairedwith a moving object, but they succeeded when the syllables were paired with a staticcheckerboard pattern. The authors argue that differences in the task demands and overallprocessing load in these two tasks are important in understanding these differences inperception within the native language. Research on infants being raised bilingually also suggests a complex pattern fornative phonetic development. Bosch and Sebastián-Gallés (2003a) studied infantsat different ages from three language groups (Spanish, Catalan, and bilingual Spanish/Catalan). They examined their perception of the /ε–e/ vowel contrast which is phonemicin Catalan but not in Spanish and is difficult for monolingual speakers of Spanish todiscriminate (Pallier, Bosch, & Sebastián, 1997). All three infant groups discriminatedthis contrast at 4 months of age. As expected, an attenuation of perception with age wasobserved for the Spanish infants but not for the Catalan infants. The bilingual groupshowed a U-shaped developmental pattern whereby discrimination declined between 4and 8 months of age, but then improved to the level of native monolingual Catalaninfants at 12 months. A U-shape developmental pattern was reported in two subsequentstudies that examined consonant perception in bilingual and monolingual infants (Bosch& Sebastián-Gallés, 2003b; Burns, Werker, & McVie, 2003). Note that this U-shape pattern was observed when bilingual infants were tested oncontrasts that have a conflicting status across their two languages – native in one of thelanguages and non-native in the other. At present it is not known why perception ofsuch contrasts reveals a temporary weak spot in the bilingual infant’s emerging nativelanguage filter(s). The conflicting functional status, reduced exposure to such contrastsin the bilingual input, or both factors, may contribute to the U-shape pattern. Alterna-tively, the bilingual infant may construe the perceptual task differently than the mono-lingual infant. (See Werker, Hall, & Fais, 2004, for a discussion of U-shape developmental 165. Infants’ Speech Perception and Production 157functions.) For now, these data pose a challenge and an opportunity for futureresearch. Overall, the developmental course that unfolds as infants begin tuning into the pho- netic structure of their native language is not uniform. Diverse developmental patternsevident in perception of native and non-native phonetic categories reveal an emergingnative language filter that operates in an active, task- specific, domain-specific, and func-tionally specific way to facilitate as well as to attenuate access to phonetic differences.To retain the filter analogy to conceptualize these assorted effects of language experience,our notion of a filter must be more sophisticated than a passive sieve. The filter evidentin the existing data is more akin to a resonator, a filtering device that can selectivelyenhance as well as selectively attenuate an input signal. Even with a more appropriateanalogy, there is no current conceptual framework that can predict when and explainhow these diverse developmental patterns arise.[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... (See Best & McRoberts, 2003, for adetailed discussion of hypotheses proposed to explain varied developmental patterns.) Abetter understanding of how language input is tied to developmental patterns is needed.There is little doubt that phonetic contrasts vary in perceptual salience in ways that areindependent of linguistic function and impact perceptual development. One challengefor future research is to provide an acceptable metric to gauge this variation in perceptualsalience and determine how it interacts with age and language experience. We must alsounderstand other factors that impact speech intake in infants and explain how thesefactors interact with language input.Variables influencing speech intake in infancyTo understand how language input shapes developmental changes in phonetic perception,we must identify factors that constrain and direct the infant’s intake of speech informa-tion. Typically, researchers present stimuli in very quiet testing environments using testparadigms that tap a perceptual function defined by the experimenter, for example dis-crimination or categorization. This approach is informative yet it tells us very little aboutwhat happens when infants encounter speech in their everyday lives. Learning throughexposure to the ambient language requires the infant to selectively attend to speech anddetect relevant patterns in the speech stream. Outside the speech lab, infants encounterspeech in presence of noise and other competing auditory and visual signals and are notexplicitly reinforced for responding to specific phonetic elements. Unlike the speech pre-sented in many laboratory studies, real language input consists of connected speech inwhich a critical feature or property of the language has not been isolated. Although verylittle is known about how infants gain access to relevant speech information “in the wild,”some studies have explored this question by assessing what infants learn from a briefexposure to a corpus of speech that has been carefully structured by the researcher. Find-ings obtained using this artificial language approach to study phonetic and phonologicaldevelopment are described in Gerken (this volume). Our discussion will focus on twoother lines of research that have explored speech intake in infants.Effects of noise and hearing loss. Infant speech perception is impacted by conditions thatalter the amount or quality of language input available, such as the presence of hearing 166. 158 Linda Polka et al.loss or the presence of background noise. Although hearing loss clearly impacts speechperception development, few systematic studies of its specific effects have been conductedwith infants. Houston, Pisoni, Iler Kirk, Ying, and Miyamoto (2003) compared atten-tion to a visual pattern, either paired with an auditory speech signal or with silence, bynormal hearing and hearing impaired infants before and after cochlear implantation(CI). Their results demonstrate a strong impact of hearing loss on infant attention tospeech. Normal hearing infants showed a strong preference for the stimulus trials witha speech signal. Hearing impaired infants showed no difference in looking time acrossspeech and silent trials before CI; a preference for speech began to emerge in some infants6 months after CI but was significantly weaker compared with normal hearing infants.Despite their[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... poor selective attention to speech, the hearing impaired infants were ableto discriminate an obvious vowel versus syllable difference (“ah” vs. “hop”) in a labora-tory test situation. However, poor attention to speech is likely to impact their processingof more subtle speech patterns in less optimal listening conditions. We found that experience with otitis media with effusion (OME) also impacts infantresponses to speech in the first year of life (Polka & Rvachew, 2005). Otitis media witheffusion causes fluid to accumulate in the middle ear and can create a mild to moderatefluctuating conductive hearing loss. To assess effects of OME on phonetic discrimina-tion, we tested 6- to 8-month-olds on a native phonetic contrast, /bu–gu/, using theconditioned headturn procedure. Tympanometry, performed after discrimination testing,revealed three groups of infants: (a) infants with middle ear effusion on test day, (b)infants who had received medical treatment for OM in the past but showing no middleear effusion on test day, and (c) infants with no history of OM and no effusion on testday. Discrimination performance was best for infants with no history of OME and worstfor infants with middle ear effusion on the day of testing. Infants with a history of OMEbut normal middle ear function on test day showed an intermediate level of discrimina-tion performance, suggesting that OME negatively impacts phonetic perception evenafter the middle ear fluid is gone. Furthermore, the poorer performance in the history-only group (b above) cannot be explained as an effect of reduced audibility and suggeststhat experience with OME affects infant attention to speech. Background noise present in many natural listening environments can also impedeaccess to relevant speech patterns. When tested in presence of noise, infants can detectphonetic differences (Nozza, Miller, Rossman, & Bond, 1991), segment words fromfluent speech (Newman & Jusczyk, 1996), and recognize their name (Newman, 2005)and their mother’s voice (Barker & Newman, 2004). However, infants need substantiallyhigher signal-to-noise ratios than adults to succeed at these tasks. Noise acts to block access to relevant sensory information and pull attention awayfrom the speech signal or critical parts of the speech signal. Although these effectsare typically inseparable in the real world, it is informative to explore how each impactsinfant perceptual processing. Moreover, as infants mature, their ability to controlattention will improve whereas the sensory impact of noise is unlikely to change. In ourlab, we tested infant phonetic perception using a distraction masker paradigm to assessthe role of selective attention independently of sensory effects of noise (Polka, Rvachew,& Molnar, submitted). We tested infant discrimination of /bu/ versus /gu/ usinga habituation procedure. To create a distractor condition a high frequency noise was 167. Infants’ Speech Perception and Production 159added to the sound file of each syllable so that it gated on and off with the onset andoffset of the syllable. The distractor noise was a recording of bird and cricket songs whosefrequencies do not overlap with the test syllables; this added signal does not make itharder to hear the syllables but it can distract infants if they[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... cannot focus their attentionwell. Infants (6–8 months) tested in quiet (i.e., unmodified syllables) performed signifi-cantly better than infants tested in the distractor condition; discrimination scores showedlittle overlap between the two groups. These findings indicate that perceiving auditoryphonetic patterns in the presence of noise poses a substantial cognitive challenge foryoung infants. Recent findings suggest that the availability of multi-modal speechinformation – an auditory-visual speech stimulus – may provide the critical supportthat makes speech perception in noise possible for infants (Hollich, Newman, &Jusczyk, 2005).Infant perceptual biases. A second way that researchers have explored how infants accessspeech information is by observing their listening preferences in test paradigms thatallow infants to control their access to speech samples. Infant listening preferences areevidence of perceptual biases that guide speech intake. Research has focused on biasesrelated to suprasegmental or indexical properties of speech such as speaker’s voice, affect,or stress (see Werker & Curtin, 2005). Recent research in our lab shows that infantsdisplay strong biases at the phonetic level as well. Our earlier cross-language studies ofinfant vowel discrimination revealed very robust and predictable directional asymmetries(see Polka & Bohn, 2003). For example, Polka and Werker (1994) found that infantswho were presented with one direction of change, /y/ to /u/, found discrimination easierthan infants presented with the same vowel pair in the reverse direction, that is, a changefrom /u/ to /y/. In this and other examples, we observed that vowel discrimination wasconsistently easier when infants were tested on a change from a less-peripheral to a more-peripheral vowel within the vowel space, suggesting that infant perception is sensitive tothe structure of the vowel space. Subsequent research using a preferential listening taskhas confirmed that these directional asymmetries reflect a strong perceptual bias forperipheral vowels (i.e., vowels closer to the corners of the F1/F2 vowel space, /i/, /a/, /u/)– the same vowels that are strongly favored in vowel inventories across languages. Vowelbiases measured in young infants are similar across infants learning different languagesbut change as language acquisition unfolds. Experiments with adults also show that thevowel biases observed in infancy have been shaped by language experience, towardoptimizing vowel processing in a specific language. We can gain many insights intospeech perception from the study of infant perceptual biases because they show us wherethe infants’ perceptual priorities lie and point to the information that they are activelyextracting from their input. Perceptual biases can also provide a relative index of per-ceptual salience and thus potentially help explain the variability in phonetic perceptionobserved in the cross-language research described above. In an effort to account for different aspects of speech perception development Werkerand Curtin (2005) recently outlined a framework that integrates a wide range of speechperception phenomena (e.g., input effects, perceptual biases, word learning) within thebroader context of language acquisition. The child’s active intake of speech informationis a prominent feature of their model, called PRIMIR (Processing[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Rich Information 168. 160 Linda Polka et al.from Multidimensional Interactive Representations). According to PRIMIR three vari-ables – initial biases, developmental level of the child, and task demands – act as dynamicfilters and work together to direct the infant’s attention to speech. Further researchinvestigating speech intake processes is needed to test the merits of the PRIMIR modeland to advance our overall knowledge of speech perception development. We have outlined some functional properties of the native language filter that isemerging in early infancy and have discussed a recent shift in research focus towardfactors that modulate speech intake in the developing infant. However, the story of infantspeech perception is incomplete until we also understand how these perceptual processesinteract with and are shaped by emerging speech production skills.Speech Production DevelopmentRecent developments in the study of infant speech production rest on the pioneeringefforts of Oller (1980), Stark (1980), and Koopmans-van Beinum and van der Stelt(1986) to describe the course of infant speech development using metrics speciallyadapted to the infant context. Oller’s (2000) comprehensive overview of this literaturecan be briefly summarized as follows: Normally developing infants universally progressthrough a series of stages of vocal development that culminate in the production ofcanonical babble, typically by 7 months of age; canonical babble is characterized by theproduction of speech-like consonant–vowel syllables, often produced as rhythmic stringsof reduplicated syllables containing stop consonants and front or central vowels; andthere is considerable continuity in the phonetic content of babble and early meaningfulspeech. More recent research has been directed at understanding the biological andexperiential influences on the form of infant vocalizations.Biological influences on early speech productionVocal tract anatomy. Major developmental changes in vocal tract shape occur shortlyafter birth, including the descent of the larynx and a sharper angle between the oral andpharyngeal cavities. A significant increase in the length of the vocal tract also occursduring the first year, with growth of the pharyngeal cavity being disproportionately largerelative to oral cavity growth throughout development (Fitch & Giedd, 1999; Kent &Vorperian, 1995; Kent, Vorperian, Gentry, & Yandell, 1999). The constraints that vocal tract morphology might place on infant speech output havebeen investigated through computer modeling by Ménard, Schwartz, and Boë (2002,2004). Although infant vocal tract anatomy does partially explain the preference for lowand front vowels, they demonstrated that the infant’s vocal tract anatomy does notprevent the production of the full range of vowels used in the ambient language.However, while it is possible to produce vowels with an infant vocal tract that are per-ceptually equivalent to adult vowel categories, in many cases the infant would need to 169. Infants’ Speech Perception and Production 161employ different articulatory gestures than the adult to achieve the same perceptualoutcome.[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Ménard et al. (2002) also identified acoustic cues to vowel identity that arevalid for the full range of vocal tract sizes. Therefore, infants can potentially perceiveequivalency between their own vowel productions and those of adult models usingthe same normalization algorithms for speech-like vocalizations produced by talkers ofall ages.Speech motor control. Infant speech production is clearly limited by immature speechmotor control abilities although the exact nature of these limitations has not beenfully determined. Indirect investigations of the infant’s ability to control the vocaltract have been conducted within the context of the frame/content theory. MacNeilage(1998) proposed that the “frame” for speech production is the repeated openingand closing of the vocal tract. The ability to modulate the “frame” to produce varied“content” is hypothesized to emerge quite late in development, with “frame dominance”persisting through the early word learning stage. Consequently, very little individualvariation in the phonetic content of babble or even early words is expected duringinfancy, within or across language groups. MacNeilage and Davis (2000) summarizeda number of studies that provide support for this hypothesis in the form of apparentlyuniversal patterns of consonant and vowel co-occurrence. However, these results arebased on phonetic transcriptions that are of questionable reliability and validity for infantspeech. Kinematic studies of infants’ articulatory movements and acoustic analyses of infants’speech output have provided a more direct and reliable picture of physiological con-straints on infant speech production (Green, Moore, Higashikawa, & Steeve, 2000;Green, Moore, & Reilly, 2002; Sussman, Duder, Dalston, & Caciatore, 1999; Sussman,Minifie, Buder, Stoel- Gammon, & Smith, 1996). Overall these data are consistent withan initial dominance of the “mandibular frame,” followed by a progressive differentiationof articulator movements. At the same time, rapid changes in speech motor controlappear to be occurring during the first 4 months of the child’s babbling experience, andlimitations on infant control of the articulatory system do not prevent the infant fromproducing a wide variety of consonant– vowel combinations even during the first year.This leads to questions about how the infant achieves this ability, given a continuouslychanging vocal tract anatomy and limited speech motor control. Guenther (1995) proposed a solution to the infant’s problem in the form of a com-putational model (Directions in Articulatory space to Velocities in Articulator space;DIVA). A fundamental characteristic of this self-organizing model is that the goal ofspeaking is considered to be the production of certain acoustic products that will beperceived by the listener as the intended target sounds. Auditory perceptual feedback isused to develop the mapping between the acoustic target and the required vocal tractconstrictions (Guenther, Hampson, & Johnson, 1998). Callan, Kent, Guenther, andVorperian (2000) demonstrated that this model adapts well to developmental changesin vocal tract size and shape. This model predicts that access to both adult and self-produced speech is critical to the development of speech motor control. Adult inputallows the infant to develop representations for language-specific[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... acoustic–phonetic 170. 162 Linda Polka et al.targets (as described above). Access to the infant’s own speech allows the child to useauditory feedback to flexibly adapt articulatory gestures to achieve the production ofthose targets.Experiential influences on speech productionEmpirical investigations have demonstrated that auditory input is crucial for normalspeech development during infancy. The emergence of the canonical babble stage isdelayed in infants with sensory–neural hearing loss (Oller & Eilers, 1988) but canonicalbabble appears in the vocalizations of hearing impaired children shortly after they receivecochlear implants (Ertmer & Mellon, 2001). These studies do not, however, illuminatethe exact nature of experiential influences on the course of prelinguistic vocal develop-ment. Oller (2000) suggests that a certain amount of auditory experience is required for“triggering the events that lead to well- formed syllable production” (p. 132). It is notclear that the specific phonetic content of the adult input is important to the process andindeed, no direct link between underlying perceptual representations and the character-istics of infant speech has been established. The dominant view posits that babbling isa strongly canalized motor behavior in which feedback of self-produced sounds servesto help the infant coordinate articulation and phonation (Koopmans-van Beinum,Clement, & van den Dikkenberg-Pot, 2001) or regulate the temporal rhythm of babble,in a manner similar to the way in which audible rattles seem to facilitate rhythmic hand-banging (Ejiri & Masataka, 2001). By contrast, the DIVA model emphasizes the role of speech input in the formation ofauditory–perceptual targets for speech output as well as the importance of auditory feed-back of the infant’s own speech for the achievement of speech motor control. This modelpredicts that the speech environment will influence the acoustic-phonetic characteristicsof the infant’s speech output from at least the beginning of the canonical babbling stage.Investigations of this hypothesis sometimes employ laboratory induced variations inspeech input to the infant but more frequently examine the impact of natural variationsin the auditory environment associated with different language groups.Laboratory investigations of the impact of speech input on speech production. Two devel-opmental changes in infant vowels occur during the first year: first, more vowels withfull oral resonance are produced in comparison to vowels with nasal resonance; subse-quently, the infant’s vowel repertoire expands to include the point vowels (/i/, /a/, /u/)in addition to the predominant low and central vowels. Two laboratory studies haveshown that auditory input plays a role in these developments. An elegant series of studies was conducted by Bloom (Bloom, 1988; Bloom, Russell,& Wassenberg, 1987) in which both the timing and content of speech input to 3-month-old infants were varied systematically in the laboratory. Infants produced a higher pro-portion of more speech-like utterances involving full resonance when they receivedspeech input (compared with non-speech vocal input), and this effect was enhancedwhen the input was provided contingent upon the infants’ vocalizations.[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... 171. Infants’ Speech Perception and Production 163 In a study focused on the specific phonetic content of the adult input and the infantspeech output, Kuhl and Meltzoff (1996) presented one of three point vowels to infantsaged 12, 16, and 20 weeks. Perceptual and acoustic analyses showed that the infantsshifted their vowel production to better match the target vowel category, even at theyoungest age. Kuhl and Meltzoff speculated that this matching-to-target phenomenon,combined with the shift to language-specific perceptual processing, underlies develop-mental changes in acoustic characteristics of infant vowel production.Cross-linguistic investigations of the impact of speech input on speech production. Descrip-tions of the prosodic characteristics of infant speech also suggest that environmentalinput impacts infant speech output. For example, French-learning infants’ babble canbe differentiated from the babble of infants learning other languages with respect tointonational contours, syllable structure, and number of syllables per utterance (Hallé,de Boysson-Bardies, & Vihman, 1991; Levitt & Utman, 1992; Whalen, Levitt, & Wang,1992). Maneva and Genesee (2002) recently reported similar findings for a single childlearning both English and French, recorded separately with his English-speaking motheror his French-speaking father. Unfortunately, these studies involved very small samplesand are largely based on phonetic transcriptions of the infants’ speech; therefore theexistence of “babbling drift,” in which infant babble “drifts” toward the ambient lan-guage in its sound properties, is not universally accepted (Oller, 2000). In our lab, we have been describing the acoustic characteristics of vowels produced byinfants learning Canadian English or Canadian French in a cross-sectional study (Rvachew,Mattock, Polka, & Ménard, 2006). Figure 8.1 shows the mean first (F1) and second (F2)formant frequencies of vowels produced by the 42 infants that we have recorded thus far.The language groups do not differ with respect to F1 or F2 frequency at 8 months ofage. At 10 months of age the groups differ in a manner similar to that reported by deBoysson-Bardies, Hallé, Sagart, and Durand (1989), with the mean F2 being significantlylower for French-learning than for English-learning infants. Between 10 and 18 monthsof age, the mean F2 for the English-learning group falls below that of the French-learninggroup which shows a rising F2 during this period. The result is a significant interactionbetween age and language group for F2 [F (3,42) = 3.68, p = .02]. These data do not represent a pattern of initial overlap of the vowel spaces followedby a linear divergence of the two groups’ “average vowel” with advancing age, as we hadexpected. Rather, the patterns of developmental change and cross-linguistic differencessuggest a more complex and dynamic situation in which the child’s ability to attend toand reproduce specific features of the native language vowel space shifts with age.However, it is difficult to interpret these findings because it is not clear what the lan-guage-specific targets are for the infant vowel space. The vowels produced by infants arenecessarily described in terms of the mean and dispersion of formant values for the entirevowel space while the adult input is universally presented as formant values for[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... specificvowels produced in an adult-directed fashion. For example, Escudero and Polka (2003)have shown that there are acoustic differences among Canadian English and CanadianFrench vowels in adult speech, even for shared point vowels. However, these data cannotbe used to predict developmental changes in the infant’s mean vowels because the adultvowel space is not described using the same method that is used for the description of 172. 164 Linda Polka et al. 3000 French 2800 English 2600 Mean F2 2400 2200 2000 1800 8 months 10 months 12–15 16–18 months months 1100 French 1000 English 900 Mean F1 800 700 600 500 400 8 months 10 months 12–15 16–18 months monthsFigure 8.1 Mean second formant frequencies (F2; top panel) and first formant frequencies (F1;bottom panel) for infants learning Canadian English (white bars) or Canadian French (speckledbars), by age group. Error bars represent standard error of the mean.infant vowels. Furthermore, it is clear that adult-directed speech is significantly differentfrom infant-directed speech and this is likely to have implications for the acoustics ofthe vowels that infants hear (e.g., see Kuhl et al., 1997). Studies in which the speechaddressed to the infant and the speech produced by the infant are recorded in the samecontext and described using the same procedures are required. In addition to establishing the characteristics of the input vowel space from theinfant’s perspective, we need to have a better understanding of how the infant processesthat input if we are to predict patterns of developmental change in speech production.For example, the non-linear changes in infant vowel characteristics that are shown inFigure 8.1 might be explained by age-related changes in infant attention to specificproperties of the adult input and the infant’s own speech. As an example, one charac-teristic of the French vowel space is the presence of rounded vowels that are cued by acomplex integration of the relatively low amplitude second, third, and fourth formantfrequencies (Ménard et al., 2004). It is possible that the cue for lip rounding has dimin-ished salience in the child’s own speech at a young age. This cue may become available 173. Infants’ Speech Perception and Production 165to the child later in the infant period, as a function of increased exposure, changes invocal tract and laryngeal anatomy, and improved control of lip movements. If the infantthen begins to try to manipulate the cue in his or her own speech output, this couldcause a non-linear shift in the child’s vowel space. However, this hypothesis cannot betested without more information about the child’s ability to attend to and process dif-ferent acoustic cues in infant and adult speech. This brings us to the issue of infantintake of speech input.The impact of variations in speech intake. The infant’s ability to receive speech input isobviously impacted by the integrity of the child’s auditory system, and hearing impair-ment has a clear impact on infant speech production as described above. To date, almostno research has considered the potential[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... role of the infant’s active efforts to listen to thespeech input and select the relevant information. It is now well known that infants listen preferentially to certain types of speech butthe developmental implications of variations in selective attention are not clear. Vihmanand Nakai (2003) reported a negative correlation between listening preferences for andproductive use of a given consonant. The results are interpreted within the context of thearticulatory filter hypothesis in which the physical act of producing a given speech soundimpacts on the child’s perception of speech and helps the child to develop “more robustlexical and phonological representations” (p. 1017). Vihman (2002) further speculatesthat articulation of adult-like syllables activates neurons that are active when performingor observing an action (i.e., mirror neurons; see Vihman for more detail). This hypothesispresumes that speech perception and speech production development are independentduring the first year of life. The onset of canonical babble is attributed to “maturation”and “rhythmic motoric advances” while the child’s implicit perceptual knowledge ofnative language sound categories is attributed to “passive intake”; when the mirror neuronsare activated and perception and production become integrated the infants “lay downphonological representations at a new level” (p. 1017; Vihman & Nakai, 2003). Note that this account of early phonetic development is based upon a correlationalstudy in which it is impossible to determine the direction of causality. The finding thatthe infants preferred to listen to consonants that they were less likely to produce is difficultto interpret. However, the focus on the child’s active intake of information from the lan-guage environment is extremely important, and there is a clear need for more studies inwhich the relationship between selective attention and speech production is explored. Interactions between the integrity of the auditory mechanism and the developmentof selective attention have been proposed, as discussed above in relation to the reducedattention to speech observed in recipients of cochlear implants (Houston et al., 2003).In particular, selective attending deficits may explain the impact of otitis media on lan-guage development. The impact is small and the clinical significance of otitis media-related language delay has been questioned (e.g., Paradise et al., 2000). However, thehearing loss associated with otitis media is so subtle and transitory that any measurableimpact on language development is surprising. And yet these impacts can be observedeven during the first year of life in the form of slower emergence of the canonical babblestage and a restricted vowel space (Rvachew, Slawinski, Williams, & Green, 1996, 1999).Current hypotheses about the way in which otitis media impacts on speech development 174. 166 Linda Polka et al.recognize that the issue is not simply one of signal audibility. The fluctuating nature ofthe hearing deficit may make it difficult for the child to discover regularities in the speechinput and lead the child to become less attentive to speech (Feagans, 1986; Mody,Schwartz, Gravel, & Ruben, 1999). Furthermore, 12-month-olds with chronic otitismedia experience fewer episodes of joint attention with their parents than children[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... withnormal hearing (Yont, Snow, & Vernon-Feagans, 2003). Thus it appears that the earlyvocalizations of children with early-onset otitis media are determined by a complexinteraction of biological factors (integrity of the peripheral auditory system), input factors(quality of input provided by parents), and intake factors (child attention to speech andability to engage in joint attention routines).Speech Perception and Speech Production Development:Challenges for the FutureThe study of infant phonetic development has been marked by tension between compet-ing perspectives about the goal of the enterprise itself, the purported basis for develop-mental change during infancy, and the significance of these changes for later languageacquisition. Specifically, some researchers have focused on the universal characteristicsof the infant’s phonetic abilities while others have investigated individual variation inphonetic development. Developmental changes in language acquisition are assumed bysome to emerge directly from the maturation of certain neurological, physiological, andanatomical structures, while others privilege the infant’s active efforts to learn from theinput. Although phonetic development during infancy is considered to be an importantfoundation for later language learning, the nature of its relation to later phonological,lexical, and syntactic development remains a question of debate. Competing perspectives on the relation between perception and production processesin early development were labeled by de Boysson- Bardies et al. (1989) as the independencehypothesis and the interactional hypothesis. From the independence perspective, speechperception and production skills develop independently during the first year, with anycontinuity between infant phonetic skills and later language development viewed as afunction of common biological constraints on perceptual and articulatory performanceduring the prelinguistic and early linguistic stages. For example, Kent and Miolo (1994)suggested that speech perception and speech production “may have somewhat differentcourses of development, but they are ultimately integrated in spoken language compe-tence” (p. 304). Locke (1994) proposed independent neurological mechanisms andmaturational timecourses to account for early phonetic development and phonologicaldevelopment after the first 50 word stage. From the interactional perspective, the child’s developing knowledge of the perceptualand articulatory characteristics of native language phonetic categories is seen as inte-grated from the beginning. For example, de Boysson-Bardies et al. (1989) stated that“articulatory procedures that are mastered step by step are oriented by auditory configu-rations” (p. 2). This perspective is associated with strong claims for continuity betweenphonetic development during this first year and language development during the second 175. Infants’ Speech Perception and Production 167year, as exemplified by Tsao, Liu, and Kuhl’s (2004) assertion that “phonetic perceptionplays a critical role in the early phases of language acquisition” (p. 1082). While the research reviewed here has not yet provided a resolution to this debate,recent changes in theoretical perspectives, methodological tools, and research[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... approachespromise considerable advances in our understanding of how perception and productionmight interact during the first year to produce the emergence of language in the secondyear of life. The DIVA model provides a theoretical account of how these domains influ-ence each other throughout the lifespan. The PRIMIR model suggests that apparentdiscontinuities in infant performance may actually be indicative of a fundamental under-lying continuity from the prelinguistic to the linguistic phases of development. Newresearch tools provide us with the opportunity to assess the merits of these theories.Investigation of individual differences in babbling after taking into account universalbiomechanical constraints has been facilitated by a number of new technologies (e.g.,magnetic resonance imaging of the vocal tract, computational modeling of the impactof vocal tract development on speech output, and kinematic and acoustic descriptions ofinfant speech production). Infant speech perception research has in the past tapped intonatural variation in language input via cross-linguistic comparisons to explore linguisticinfluences. Researchers are beginning to actively control and manipulate language experi-ence. For example, Kuhl, Tsao, and Liu (2003) manipulated different components of anatural language setting to show the impact of a live social context on phonetic learning.Given the wide acceptance of highly controlled artificial language paradigms, cross-linguistic studies implementing this approach will soon provide insights into the lan- guage-specificity (or lack thereof) of infant on-line speech processing. Along with thesenew methods there is a growing interest in examining individual differences in perceptualresponding under conditions that are more akin to the infant’s natural environment. Together these new technologies make it possible for researchers, currently workingseparately in the domains of speech perception or speech production, to come togetherto investigate the relationship between individual variations in speech perception andproduction performance in the same infants. The DIVA and PRIMIR models suggestsome specific directions for future research. The PRIMIR model predicts that there willbe individual differences and developmental changes in the intake of specific aspects ofthe rich and multidimensional input available to the infant. Ongoing efforts to under-stand the infant’s role in the selection of input are extremely important. The DIVA modelshows that a critical but as yet unexplored aspect of this input is the infant’s own speech.Collaborative and interdisciplinary research is needed to focus research on the interplaybetween perception and production processes within the developing child. The resultswill have a profound impact on our understanding of early language development.Notes1 Note that no studies of lexical tone perception have been conducted with infants younger than 6 or older than 9 months so it is not known whether the perceptual decline actually occurs earlier in development or continues later.2 The lower discrimination levels in the Polka et al. study are an exception. 176. 168 Linda Polka et al.ReferencesBarker, B. A., & Newman, R. S. (2004). Listen to your mother! The role of talker familiarity in infant streaming.[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Cognition, 94, B45–B53.Best, C. T. (1991, April). Phonetic influences on the perception of non-native speech contrasts by 6–8 and 10–12-month-olds. Paper presented at the Meeting of Society for Research in Child Development, Seattle, WA.Best, C. T. (1995). A direct-realist view of cross- language speech perception. In W. Strange (Ed.), Speech perception and early linguistic experience (pp. 171–206). Baltimore, MD: York Press.Best, C., & McRoberts, G. W. (2003). Infant perception of non-native consonant contrasts that adults assimilate in different ways. Language and Speech, 46, 183–216.Best, C. T., McRoberts, G. W., Goodell, E., Wormer, J. S., Insabella, G., Kim, P., et al. (1990, April). Infant and adult perception of non-native speech contrasts differing in relation to the lis- tener’s native phonology. Paper presented at the Meeting of the International Conference on Infant Studies, Montreal.Best, C. T., McRoberts, G., LaFleur, R., & Silver-Isenstadt, J. (1995). Divergent developmental patterns for infants’ perception of two nonnative consonant contrasts. Infant Behavior and Development, 18, 339– 350.Best, C. T., McRoberts, G. W., & Sithole, N. M. (1988). Examination of perceptual reorganiza- tion for nonnative speech contrasts: Zulu click discrimination by English-speaking adults and infants. Journal of Experimental Psychology: Human Perception and Performance, 14, 345– 360.Best, C., Singh, L., Bouchard, J., Connelly, G., Cook, A., & Faber, A. (1997, April). Develop- mental changes in infants’ discrimination of non-native vowels that assimilate to two native catego- ries. Paper presented at the Meeting of Society of Research in Child Development.Bloom, K. (1988). Quality of adult vocalizations affects the quality of infant vocalizations. Journal of Child Language, 15, 469–480.Bloom, K., Russell, A., & Wassenberg, K. (1987). Turn taking affects the quality of infant vocalizations. Journal of Child Language, 14, 211–227.Bosch, L., & Sebastián-Gallés, N. (2003a). Simultaneous bilingualism and the perception of a language-specific vowel contrast in the first year of life. Language and Speech, 46, 217–243.Bosch, L., & Sebastián-Gallés, N. (2003b). Language experience and the perception of a voic- ing contrast in fricatives: Infant and adult data. Paper presented at the 15th International Conference of the Phonetic Sciences, Barcelona, Spain.Burns, T. C., Werker, J. F., & McVie, K. (2003). Development of phonetic categories in infants raised in bilingual and monolingual environments. In B. Beachley, A. Brown, & F. Conlin (Eds.), Proceedings of the 27th Annual Boston University Conference on Language Development. Somerville, MA: Cascadilla Press.Callan, D. E., Kent, R. D., Guenther, F. H., & Vorperian, H. K. (2000). An auditory-feedback- based neural network model of speech production that is robust to developmental changes in the size and shape of the articulatory system. Journal of Speech, Language, and Hearing Research, 43, 721–738.De Boysson-Bardies, B., Hallé, P., Sagart, L., & Durand, C. (1989). A crosslinguistic investiga- tion of vowel formants in babbling. Journal of Child Language, 16, 1–17.Eilers, R. E., Gavin, W. J., & Oller, D. K. (1982). Cross-linguistic perception in infancy: Early effects of linguistic experience. Journal of Child Language, 9, 289–302.Ejiri, K., &[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Masataka, N. (2001). Co-occurrence of preverbal vocal behavior and motor action in early infancy. Developmental Science, 4, 40–48. 177. Infants’ Speech Perception and Production 169Ertmer, D. J., & Mellon, J. A. (2001). Beginning to talk at 20 months: Early vocal development in a young cochlear implant recipient. Journal of Speech, Language, and Hearing Research, 44, 192–206.Escudero, P., & Polka, L. (2003). A cross-language study of vowel categorization and vowel acoustics: Canadian English versus Canadian French. Paper presented at Paper presented at the 15th International Conference of the Phonetic Sciences, Barcelona, Spain.Feagans, L. (1986). Otitis media: A model for long-term effects with implications for interven- tion. In J. F. Kavanagh (Ed.), Otitis media and child development (pp. 192–208). Parkton, MD: York Press.Fitch, W. T., & Giedd, J. (1999). Morphology and development of the human vocal tract: A study using magnetic resonance imaging. Journal of the Acoustical Society of America, 106, 1511–1522.Green, J. R., Moore, C. A., Higashikawa, M., & Steeve, R. W. (2000). The physiologic develop- ment of speech motor control: Lip and jaw coordination. Journal of Speech, Language, and Hearing Research, 43, 239–255.Green, J. R., Moore, C. A., & Reilly, K. J. (2002). The sequential development of jaw and lip control for speech. Journal of Speech, Language, and Hearing Research, 45, 66–79.Guenther, F. H. (1995). Speech sound acquisition, coarticulation, and rate effects in a neural network model of speech production. Psychological Review, 102, 594–621.Guenther, F. H., Hampson, M., & Johnson, D. (1998). A theoretical investigation of reference frames for the planning of speech movements. Psychological Review, 105, 611– 633.Hallé, P. A., de Boysson-Bardies, B., & Vihman, M. M. (1991). Beginnings of prosodic organiza- tion: Intonation and duration patterns of disyllables produced by Japanese and French infants. Language and Speech, 34, 299–318.Hollich, G., Newman, R. S., & Jusczyk, P. W. (2005). Infant’s use of synchronized visual infor- mation to separate streams of speech. Child Development, 76, 1–16.Houston, D. M., Pisoni, D., Iler Kirk, K., Ying, E. A., & Miyamoto, R. T. (2003). Speech per- ception skills of deaf infants following cochlear implantation: a first report. International Journal of Pediatric Otorhinolaryngology, 67, 479–495.Kent, R. D., & Miolo, G. (1994). Phonetic abilities in the first year of life. In P. Fletcher & B. MacWhinney (Eds.), Handbook of child language (pp. 303–334). Oxford: Blackwell.Kent, R. D., & Vorperian, H. K. (1995). Anatomic development of the craniofacial-oral laryngeal systems: A review. Journal of Medical Speech-Language Pathology, 3, 145–190.Kent, R. D., Vorperian, H. K., Gentry, L. R., & Yandell, B. S. (1999). Magnetic reson- ance imaging procedures to study the concurrent anatomic development of vocal tract structures: preliminary results. International Journal of Pediatric Otorhinolaryngology, 49, 197– 206.Koopmans-van Beinum, F. J., Clement, C. J., & van den Dikkenberg-Pot, I. (2001). Babbling and the lack of auditory speech perception: a matter of coordination? Developmental Science, 4, 61–70.Koopmans-van Beinum, F. J., & van der Stelt, J. M. (1986). Early stages in the development of speech[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... movements. In B. Lindblom & R. Zetterstrom (Eds.), Precursors of early speech (pp. 37–50). New York: Stockton Press, Inc.Kuhl, P. K., Andruski, J. E., Chistovich, I. A., Kozhevnikova, E. V., Ryskina, V. L., Stolyarova, E. I., et al. (1997). Cross-language analysis of phonetic units in language addressed to infants. Science, 277, 684–686.Kuhl, P. K., & Meltzoff, A. N. (1996). Infant vocalizations in response to speech: Vocal imitation and developmental change. Journal of the Acoustical Society of America, 100, 2425–2438. 178. 170 Linda Polka et al.Kuhl, P. K., Tsao, F.-M., & Liu, H.-M. (2003). Foreign-language experience in infancy: Effects of short-term exposure and social interaction on phonetic learning. Proceedings of the National Academy of Sciences, 100, 9096–9101.Levitt, A. G., & Utman, J. A. (1992). From babbling towards the sound systems of English and French: A longitudinal two-case study. Journal of Child Language, 19, 19–49.Locke, J. L. (1994). Development of the capacity for spoken language. In P. Fletcher & B. MacWhinney (Eds.), Handbook of child language (pp. 278–302). Oxford: Blackwell.MacNeilage, P. F. (1998). The frame/content theory of evolution of speech production. Behavioral and Brain Sciences, 21, 499–546.MacNeilage, P. F., & Davis, B. L. (2000). On the origin of internal structure of word forms. Science, 288, 527–531.Maneva, B., & Genesee, F. (2002). Bilingual babbling: Evidence for language differentiation in dual language acquisition. In B. Skarbela et al. (Eds.), Boston University Conference on Language Development 26 Proceedings (pp. 383–392). Somerville, MA: Cascadilla Press.Mattock, K., & Burnham, D. (2006). Chinese and English infants’ tone perception: Evidence for perceptual reorganization. Infancy, 10.Ménard, L., Schwartz, J., & Boë, L. (2002). Auditory normalization of French vowels synthesized by an articulatory model simulating growth from birth to adulthood. Journal of the Acoustical Society of America, 111, 1892– 1905.Ménard, L., Schwartz, J., & Boë, L. (2004). Role of vocal tract morphology in speech develop- ment: Perceptual targets and sensorimotor maps for synthesized vowels from birth to adult- hood. Journal of Speech, Language, and Hearing Research, 47, 1059–1080.Miyawaki, K., Strange, W., Verbrugge, R., Liberman, A., Jenkins, J., & Fujimura, O. (1975). An effect of language experience: The discrimination of /r/ and /l/ by native speakers of Japa- nese and English. Perception and Psychophysics, 18, 331–340.Mody, M., Schwartz, R. G., Gravel, J. S., & Ruben, R. J. (1999). Speech perception and verbal memory in children with and without histories of otitis media. Journal of Speech, Language, and Hearing Research, 42, 1069– 1079.Newman, R. S. (2005). The cocktail party effect revisited: Listening to one’s own name in noise. Developmental Psychology, 41, 352–362.Newman, R. S., & Jusczyk, P. W. (1996). The cocktail party effect in infants. Perception and Psychophysics, 58, 1145–1156.Nozza, R. J., Miller, S. L., Rossman, R. N. F., & Bond, L. C. (1991). Reliability and validity of infant speech-sound discrimination-in-noise thresholds. Journal of Speech and Hearing Research, 34, 643–650.Oller, D. K. (1980). The emergence of the sounds of speech in infancy. In G. H. Yeni-Komshian, J. F. Kavanagh, & C. A. Ferguson (Eds.),[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Child phonology: Vol. 1 (pp. 93–112). New York: Academic Press.Oller, D. K. (2000). The emergence of the speech capacity. Mahwah, NJ: Lawrence Erlbaum Associates.Oller, D. K., & Eilers, R. E. (1988). The role of audition in infant babbling. Child Development, 59, 441–449.Pallier, C., Bosch, L., & Sebastián, N. (1997). A limit on behavioral plasticity in vowel acquisi- tion. Cognition, 64, B9–B17.Paradise, J. L., Dollaghan, C. A., Campbell, T. F., Feldman, H. M., Bernard, B. S., Colborn, K., et al. (2000). Language, speech sound production, and cognition in three-year-old children in relation to otitis media in their first three years of life. Pediatrics, 105, 1119–1130.Pegg, J. E., & Werker, J. F. (1997). Adult and infant perception of two English phones. Journal of the Acoustical Society of America, 102, 3742–3753. 179. Infants’ Speech Perception and Production 171Polka, L., & Bohn, O. S. (1996). A cross-language comparison of vowel perception in English- learning and German-learning infants. Journal of the Acoustical Society of America, 100, 577–592.Polka, L., & Bohn, O.-S. (2003). Asymmetries in vowel perception. Speech Communication, 41, 221–231.Polka, L., Colantonio, C., & Sundara, M. (2001). A cross-language comparison of /d/–/ð/ per- ception: Evidence for a new developmental pattern. Journal of the Acoustical Society of America, 109, 2190–2201.Polka, L., & Rvachew, S. (2005). The impact of otitis media with effusion on infant phonetic perception. Infancy, 8, 101– 117.Polka, L., Rvachew, S., & Molnar, M. (submitted). The role of attention in infant speech percep- tion. Cognition.Polka, L., & Werker, J. F. (1994). Developmental changes in perception of nonnative vowel con- trasts. Journal of Experimental Psychology: Human Perception and Performance, 20, 421– 435.Rvachew, S., Mattock, K., Polka, L., & Ménard, L. (2006). Developmental and cross- linguistic variation in the infant vowel space: the case of Canadian English and Canadian French. Journal of the Acoustical Society of America, 120.Rvachew, S., Slawinski, E. B., Williams, M., & Green, C. (1996). Formant frequencies of vowels produced by infants with and without early onset otitis media. Canadian Acoustics, 24, 19–28.Rvachew, S., Slawinski, E. B., Williams, M., & Green, C. L. (1999). The impact of early onset otitis media on babbling and early language development. Journal of the Acoustical Society of America, 105, 467–475.Stager, C. L., & Werker, J. (1997). Infants listen for more phonetic detail in speech perception than in word-learning tasks. Nature, 388, 381–382.Stark, R. (1980). Stages of speech development in the first year of life. In G. H. Yeni-Komshian, J. F. Kavanagh, & C. A. Ferguson (Eds.), Child phonology: Vol. 1 (pp. 73–92). New York: Academic Press.Sundara, M., Polka, L., & Genesee, G. (2006). Language experience facilitates discrimination of /d–ð/ in monolingual and bilingual acquisition of English. Cognition, 100, 369–388.Sussman, H. M., Duder, C., Dalston, E., & Caciatore, A. (1999). An acoustic analysis of the development of CV coarticulation: A case study. Journal of Speech, Language, and Hearing Research, 42, 1080–1096.Sussman, H. M., Minifie, F. D., Buder, E. H., Stoel- Gammon, C., & Smith, J. (1996). Conso- nant–vowel interdependencies in babbling and early words: Preliminary examination of a locus equation[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... approach. Journal of Speech, Language, and Hearing Research, 39, 424– 433.Tsao, F., Liu, H.-M., & Kuhl, P. K. (2004). Speech perception in infancy predicts language devel- opment in the second year of life: A longitudinal study. Child Development, 75, 1067–1084.Tsao, F., Liu, H., Kuhl, P. K., & Tseng, C. (2000). Perceptual discrimination of Mandarin frica- tive–affricate contrast by English-learning and Mandarin-learning infants. Paper presented at the International Conference on Infant Studies, Brighton, UK.Tsushima, T., Takizawa, O., Sasaki, M., Shiraki, S., Nishi, K., Kohno, M., et al. (1994). Dis- crimination of English /r–l/ and /w–y/ by Japanese infants at 6–12 months: Language-specific developmental changes in speech perception abilities. Paper presented at the International Confer- ence on Spoken Language Processing.Vihman, M. M. (2002). The role of mirror neurons in the ontogeny of speech. In M. I. Stamenov & V. Gallese (Eds.), Mirror neurons and the evolution of brain and language (pp. 305–314). Advances in Consciousness Research: Vol. 42. Amsterdam: John Benjamins Publishing Company. 180. 172 Linda Polka et al.Vihman, M. M., & Nakai, S. (2003). Experimental evidence for an effect of vocal experience on infant speech perception. In M. J. Sole, D. Recasens, & J. Romero (Eds.), Proceedings of the 15th International Congress of Phonetic Science, Barcelona (pp. 1017–1020). Universitat Autonoma de Barcelona.Werker, J. F., & Curtin, S. (2005). PRIMIR: A developmental framework of infant speech pro- cessing. Language Development and Learning, 1, 197–234.Werker, J. F., Gilbert, J. H. V., Humphrey, K., & Tees, R. C. (1981). Developmental aspects of cross- language speech perception. Child Development, 52, 349–355.Werker, J. F., Hall, G., & Fais, L. (2004). Reconstructing U-shaped functions. Journal of Cognition and Development, 5, 147–151.Werker, J. F., & Logan, J. (1985). Cross-language evidence for three factors in speech perception. Perception and Psychophysics, 37, 35–44.Werker, J. F., & Tees, R. C. (1983). Developmental changes across childhood in the perception of nonnative speech sounds. Canadian Journal of Psychology, 37, 278–286.Werker, J. F., & Tees, R. C. (1984a). Cross-language speech perception: Evidence for perceptual reorganisation in the first year of life. Infant Behavior and Development, 7, 49–63.Werker, J. F., & Tees, R. C. (1984b). Phonemic and phonetic factors in adult cross-language speech perception. Journal of the Acoustical Society of America, 75, 1866–1878.Whalen, D. H., Levitt, A. G., & Wang, Q. (1992). Intonational differences between the redu- plicative babbling of French- and English-learning infants. Journal of Child Language, 18, 501–516.Yont, K. M., Snow, C. E., & Vernon-Feagans, L. (2003). Is chronic otitis media associated with differences in parental input at 12 months of age? An analysis of joint attention and directives. Applied Psycholinguistics, 24, 581–602. 181. 9Acquiring Linguistic StructureLouAnn GerkenThis chapter is an overview of what scientists currently know about human sensitivityto linguistic form during infancy. We can think of language form at two levels: a sub-[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... meaning level that includes the sounds that combine to make words in a spoken lan-guage, and a meaning level that includes words and phrases that combine to makesentences. The ability to generate new combinations at both of these levels is what giveshuman language its infinite creativity. It is standard when discussing linguistic form tosubdivide the territory into linguistic categories of various sorts (e.g., the phonologicalcategory “stop consonants” or the syntactic category “nouns”) and rules, principles, orstatistical regularities describing the typical ways in which these categories are combined(e.g., a stop cannot follow a liquid in word initial position in English). The inclusion of such a chapter in a book on language development would probablynot even have been considered 20 years ago. Why has the topic of infants’ sensitivity tolinguistic form become one of so much interest? I can identify two reasons, one meth-odological and one theoretical. The first is that, largely for technological reasons, theearliest studies of infant sensitivity to language asked questions about young learners’ability to discriminate acoustic–phonetic forms without any reference field (e.g., Eimas,Siqueland, Jusczyk, & Vigorrito, 1971; Werker & Tees, 1984). Later studies examininginfants’ sensitivity to linguistic units larger than individual speech sound or syllablescontinued to use discrimination, not association with reference, as the method of choice(e.g., Hirsh-Pasek et al., 1987; Jusczyk & Aslin, 1995). As the questions asked of infantsusing form discrimination measures became more and more linguistically sophisticated,it became clear that sensitivity to form may precede in many respects the ability to mapforms to meanings (Gómez & Gerken, 2000; Naigles, 2002; see Naigles & Swensen,this volume). A second reason for the field’s interest in infants’ sensitivity to linguistic form concernsthe debate about whether language is learnable using a set of general purpose learning 182. 174 LouAnn Gerkenmechanisms, or whether we must posit strong innate constraints on the language acquisi-tion process (e.g., Chomsky, 1981; see Saffran & Thiessen, this volume). This debatehas often focused on linguistic form, particularly syntactic form. An argument favoringnativist views of language development is that any set of data can potentially give riseto an infinite number of generalizations. How can a learner be sure that she is makingthe correct generalization, given the data? The nativist solution to this question is toposit that learners are born strongly constrained to consider only a very restricted set ofpossible generalizations. In the limit, a single input datum might trigger the correctgeneralization in a particular linguistic domain (e.g., whether sentences require overtsubjects; Hyams, 1986). Contrary to such views, the recent research on infants’ sensitivity to linguistic formhints at the possibility that, given a reasonable subset of the input data, infants arecapable of converging on the appropriate linguistic generalizations, possibly using generalpurpose learning mechanisms coupled with general purpose perceptual/conceptual con-straints. The logical observation that any set of input admits multiple possible generaliza-tions which somehow must be constrained can be kept distinct from an empirical claimabout the nature of[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... the input – the “poverty of the stimulus” argument. This argumentstates that certain critical types of linguistic data are so rare that learners are not exposedto them early in language development (see Pullum & Scholz, 2002 and responses).Nevertheless, nativists argue, children have knowledge of the formal principles underly-ing the putatively unheard data. I return briefly to the poverty of the stimulus argumentin the final section. The data that I present in this chapter all come from experiments in which 6- to 18-month-old infants are tested on their preference for one auditory stimulus type versusanother, with preference defined as greater attention to one stimulus type than the otherover multiple trials in a controlled setting. In some studies, infants are tested on theirability to discriminate two types of stimuli based on existing knowledge they had whenentering the laboratory. In other studies, infants are familiarized with new auditorystimuli and then tested to determine whether they can discriminate the newly familiar-ized stimuli from very similar stimuli. The question each researcher is asking is whetherinfants can discriminate two types of stimuli based on form alone, without assessingtheir interpretation of utterances. The careful reader will note that in some studies dis-crimination is reflected in greater attention to familiar forms, while in other studies thereis greater attention to novel forms. Which type of preference is observed in which typeof experiment may be related to how well infants were able to encode the relevant prop-erties of the stimulus before testing, which itself is probably affected by the age of theinfant, the length of exposure before testing, the complexity of the stimulus, and thecomplexity of the testing environment. At this point in the development of the field,researchers focus on whether or not infants demonstrate significant discrimination,regardless of the direction. The next two sections address what is known about infants’ sensitivity to phonologicaland syntactic form, in terms of categories and combinatorial regularities. The finalsection addresses what, if anything, the findings of infant sensitivity to linguistic formtell us about the nature of language development. 183. Acquiring Linguistic Structure 175Sensitivity to Phonological FormLanguages of the world demonstrate a variety of patterns in the sounds that they use.For example, they select a subset of all possible humanly producible and perceivablesounds, and they do so in such a way that the sounds can be organized along a smallnumber of dimensions (i.e., phonetic features such as voicing). Languages also restrictwhich sounds can occur in sequence, and again, they do so based not on particularsounds, but on featurally defined sound classes. Finally, languages assign stress to sylla-bles of multi-syllabic words based on certain abstract properties, such as syllable shape(e.g., consonant–vowel–consonant) and position in a word (e.g., second to last). Below,we will consider what is known about infants’ sensitivity to information in the speechsignal relevant in each of these three areas.Sensitivity to phonetic featuresLet us begin this section by considering how infants determine what acoustic differencesare relevant in their language and which are not. We might naïvely assume that infantslose their ability to[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... discriminate sounds that are not in the input. However, such anassumption misses the point that many acoustic differences that are phonemic in onelanguage appear in another language as allophones (contextually conditioned variants)of a single phoneme. For example, English-speakers have the option of releasing or notreleasing and aspirating word final stops. Thus, English-learning infants may be exposedto both released and unreleased stops, but this phonetic difference does not affectmeaning in English. The same acoustic difference does affect meaning in Hindi. Whatcauses the English-learning infant and the Hindi-learning infant, both of whom hearvariation in aspiration in their input, to treat aspiration differently? One class of hypotheses is based on the observation that infants show a decline innon-native consonant discrimination at roughly the period of development that theybegin to recognize and produce first words (e.g., Best, 1995; Jusczyk, 1985; MacKain,1982; Werker & Pegg, 1992). Perhaps associating word forms with meanings as part ofbuilding a lexicon causes learners to focus on which aspects of form are relevant tomeaning and which are not. A potential problem with this view is that infants’ abilityto discriminate non-native vowel sounds declines at about 6 months, a time at whichword learning is not obviously underway (Kuhl et al., 1992; Polka & Werker, 1994). Ifa non-lexical mechanism for perceptual change exists for vowels, the same mechanismmay explain developmental change in consonant perception as well. Another problemwith views that depend on word learning for change in speech sound discrimination isthat infants appear to have difficulty discriminating minimal word pairs at the earlystage of word learning (Stager & Werker, 1997; Werker, Fennell, Corcoran, & Stager,2002; see Polka, Rvachew, & Mattock, this volume). For example, an infant who easilydiscriminates ba from pa might have difficulty discriminating bear from pear inthe early stages of word learning. It is difficult to see how such an infant could use 184. 176 LouAnn Gerkenword–meaning pairs to focus on voicing as an important feature of English words.Finally, even if a learner were able to use the meaning distinction between “bear” and“pear” to determine that /b/ and /p/ are distinct, this realization by itself does nothingto help them determine that the feature voicing is distinctive in English. In other words,do learners need to encounter a minimal pair that contrasts each possible pair of Englishphonemes (e.g., /b/ vs. /p/, /d/ vs. /t/, /z/ vs. /s/, etc.)? Or does determining that thephonetic feature voicing is important for distinguishing one pair of phonemes “buy” theinfant a whole set of distinctions that depend on the feature voicing? The question ofinfants’ sensitivity to phonetic features is important in the discussion in this and thefollowing two sections. Another hypothesis about the mechanism that underlies infants’ focus on the phoneticfeatures that are relevant in the target language concerns their attention to the statisticalproperties of their input (e.g., Guenther & Gjaja, 1996; Maye, Werker, & Gerken, 2002).On this view, an English-learning infant might hear a continuum of different degreesof aspiration on word final[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... stops, with most of the values clustering around a particularpoint in the acoustic distribution. That is, English-learning infants are likely to hear aunimodal distribution of aspiration. Hindi-learning infants are also likely to hear a rangeof aspiration values; however, the values should cluster around two points in the distribu-tion – one for segments in which the speaker intends aspiration and the other for inten-tionally unaspirated segments. Thus, the Hindi-learner is exposed to a bimodaldistribution of this acoustic variable. Research by Maye and colleagues suggests that even 6-month-olds respond differentlyto uni- versus bimodal distributions of speech sounds (Maye et al., 2002). Six- and 8-month-old infants were exposed for about two minutes to syllables that varied along theacoustic dimension represented by the endpoints of [d] as in day and the unaspirated [t]in stay along with filler stimuli (adult English-speakers perceive both endpoints as /d/).All infants heard all of the stimuli from an eight-token continuum. However, half of theinfants heard a stimulus set in which most tokens came from the middle of the contin-uum (tokens 4 & 5, unimodal group), while the other half heard a set in which mosttokens came from near the endpoints (tokens 2 & 7, bimodal group). During test,infants’ listening times were measured as they were exposed to trials comprising eitheran ongoing alternation between the two endpoints (tokens 1 & 8, alternating trials) ora single stimulus from the continuum repeated (tokens 1 or 8, non-alternating trials).Each trial ended when the infant stopped fi xating the visual target for a predeterminedtime. Only infants from the bimodal group responded differentially to the alternatingversus non- alternating trials. One interpretation of these findings is that exposure to a bimodal distribution helpedinfants determine that the acoustic dimension in question was potentially relevant. Bycontrast, exposure to a unimodal distribution made it more likely that infants wouldignore the same acoustic difference. These results suggest that infants are able to performsome sort of tacit descriptive statistics on acoustic input. Does this statistical analysisreveal which speech sounds are distinct from each other in a pair-by-pair fashion (e.g.,/pa/ vs. /ba/, /ta/ vs. /da/, etc.), or does it also reveal more abstract ways in which speechsounds might differ from each other (i.e., phonetic features)? Maye and Weiss (2003)found that 8-month-olds familiarized with a bimodal [d]~[t] continuum like that 185. Acquiring Linguistic Structure 177described in the preceding paragraph were able to discriminate a different, [g]~[k],continuum, which is based on the same phonetic feature. By contrast, infants exposedto a unimodal [d]~[t] continuum were not able to discriminate the [g]~[k] continuum.These data indicate that infants are able to generalize a contrast discovered via the sta- tistics over one pair of sounds to another pair of sounds differing on the same featuraldimension. It is interesting to note that adults exposed to stimuli similar to thoseemployed by Maye and Weiss failed to generalize (Maye & Gerken, 2001). However,Maye and Weiss used multiple versions of each token in the continuum, while the studywith adults did not. Therefore, we cannot determine at this point whether infants aremore adept at featural[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... generalization than adults, or if infants in the existing experimentswere presented with stimuli that better promoted feature-based generalization.Sensitivity to segment sequencesTwo lines of research suggest that infants are sensitive to segment sequences in the speechstream. The first line was begun by Peter Jusczyk and colleagues, and it demonstratesthat infants are able to discriminate words composed of sequences of segments that occurfrequently in the infants’ native language from less frequent (or entirely absent) sequences(Gerken & Zamuner, in press; Jusczyk, Friederici, Wessels, Svenkerud, & Jusczyk, 1993;Jusczyk, Luce, & Charles-Luce, 1994; Sebastián Gallés & Bosch, 2002). The secondline of research demonstrates that infants are able to learn new segment-sequencing pat-terns in a brief laboratory exposure. In one study of the latter sort, Chambers, Onishi,and Fisher (2003) familiarized 16.5-month-old infants with consonant–vowel– consonant (CVC) syllables in which particular consonants were artificially restricted toeither initial or final position (e.g., /bæp/ not /pæb/). During test, infants listened sig-nificantly longer to new syllables that violated the familiarized positional constraintsthan to new syllables that obeyed them. In this study, infants could have respondedbased on familiar segment-by- syllable position correlations (e.g., b first, p last). That is,there is no evidence that they encoded the sequence constraints in terms of features. Two similar studies suggest that infants are able to encode segment sequences in termsof featural relations. Saffran and Thiessen (2003) familiarized 9-month- olds with wordswith a consistent word-shape template. For example, in one condition of their secondexperiment, infants were familiarized with CVCCVC words which had the pattern +V,−V, +V, −V on the four consonants (e.g., /gutbap/). Infants were then tested to determineif they were able to segment from fluent speech new words that fit versus did not fit thefamiliarized pattern. The familiarization and test words were designed so that no par-ticular sequence of consonants occurred in both familiarization and test (e.g., g_tb_poccurred in familiarization but not in test, and g_kb_p occurred in test but not infamiliarization). Therefore, the influence of the familiarization phase on infants’ prefer-ence during test was probably due to word templates specified in terms of features, notspecific phonemes. A similar point is made by Seidl and Buckley (2005), who demonstrated that 9-month-olds exposed to a phonological pattern instantiated with one set of segments couldrecognize the pattern instantiated in another set of segments. In one condition of one 186. 178 LouAnn Gerkenexperiment, infants were familiarized with stimuli that exhibited the restriction thatfricatives and affricates occurred only between two vowels, and no stops occurred in thatposition (e.g., [pasat nod εt mitʃa]). During test, infants discriminated stimuli thatadhered to the restriction from stimuli that did not, even though the set of fricatives,stops, and vowels used in the test stimuli were different from those used during famil-iarization. These data, like those of Maye and Weiss (2003) and Saffran and Thiessen(2003), suggest that infants generalize about the sound properties of their language[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... basedon phonetic features.Sensitivity to properties affecting stress assignmentAs for infants’ sensitivity to stress assignment principles, research has followed a trajec-tory similar to that of explorations of sensitivity to segment sequences. Early studiesasked whether infants are sensitive to the canonical stress pattern of their language, whilelater studies asked what infants can learn about stress assignment principles in a brieflaboratory exposure. Beginning with what infants know about the stress properties oftheir own language, Jusczyk, Cutler, and Redanz (1993) demonstrated that 9-month-oldAmerican infants listen longer to disyllabic words exhibiting a trochaic pattern (strong–weak) than an iambic pattern (weak–strong). The vast majority of disyllabic words inEnglish exhibit a trochaic pattern (Cutler & Carter, 1987), and it appears that English-learning infants have noticed this statistical bias in their language (also see Echols,Crowhurst, & Childers, 1997; Thiessen & Saffran, 2003). The trochaic bias in English words can be seen to stem from a set of stress assignmentprinciples such as those in (1a– d), below (e.g., Hogg & McCully, 1987). (1) a. Stress penultimate (second to last) syllables b. Stress heavy syllables (CV with long vowel or CVC(C)(C)) c. Avoid two stressed syllables in sequence d. Alternate stress from right to left. Turk, Jusczyk, and Gerken (1995) asked whether infants were sensitive to the principlethat heavy syllables should receive stress, examining infants’ listening time to trochaicversus iambic words in which the strong syllable was light (a CVC with a short vowel).They found that syllable weight is not a necessary component of the strong–weak prefer-ence observed by Jusczyk, Cutler, et al. (1993). However, the third experiment in thepublished series, plus additional unpublished experiments, make it clear that infants aresensitive to syllable weight and to the typical patterns of heavy and light syllables thatoccur in English words. Gerken (2004) further explored infants’ sensitivity to stress assignment principles,utilizing principles and stimuli created by Guest, Dell, and Cole (2000) for a study withadults. In the infant study, 6- and 9-month-olds were familiarized with five types ofthree- to five-syllable words from one of two artificial languages that differed in mostof their stress assignment principles. No single familiarization word type exhibited allof the stress assignment principles for the language. During test, infants heard new words 187. Acquiring Linguistic Structure 179with different stress patterns from the ones heard during familiarization, although thetest words of each language were consistent with the stress assignment principles of thatlanguage. Importantly, Language 1 and Language 2 test words had the same stress pat- terns, and differed only in the placement of a heavy (CVC) syllable. For example, do-TON-re-MI-fa was a test word from Language 1, and do-RE-mi- TON-fa was a test wordfrom Language 2 (capital letters indicate stressed syllables). Nine-month-olds discrimi-nated the test words, suggesting that they were able to generalize to new words by com-bining information from the different types of words encountered during familiarization.Six-month-olds did not discriminate the test words. Can we conclude from this study that infants[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... infer stress assignment principleslike (1a–d)? One barrier to drawing such a conclusion is that the only heavy syllableused by Gerken (2004) was TON. Therefore, infants might have determined thatTON should be stressed, while not drawing the more abstract conclusion that heavysyllables should be stressed. Unpublished follow-up studies suggest a more complicatedstory. Infants failed to generalize to test stimuli with a different heavy syllable than theone heard during familiarization. However, if multiple heavy syllables were heard duringfamiliarization, infants were able to generalize to a new heavy syllable at test. Thesedata suggest that infants are not prepared to infer a principle like “stress heavy syllables”from encountering a single heavy syllable. However, they do appear to generalize basedon categories like “heavy syllable” if they hear a small number of exemplars fromthat category.Summary of sensitivity to phonological formThe studies of infants’ sensitivity to phonological form reveal at least two propertiescommon across phonological domains. First, infants demonstrate sensitivity to thesegment inventory, segment sequences, and stress properties of their native language atabout 9 months of age, although sensitivity to the vowel inventory appears somewhatearlier. Second, infants are remarkably skilled at detecting phonological patterns instimuli presented in brief laboratory visits, and they appear to be able to generalizebeyond the particular stimuli that they have encountered when given appropriate evi-dence. In the domain of segment inventories, appropriate evidence for the existence ofa featurally based segment category may be a bimodal distribution of acoustic–phonetictokens along a particular acoustic dimension, and perhaps multiple instances of eachtoken. In the domain of segment sequences, there has not been a systematic explorationof what evidence is required for featurally based generalization. For example, althoughit seems unlikely that infants familiarized with a set of pVsVC tokens would show evi-dence of having induced a stop– V–fricative–VC pattern, we do not yet know the limitson infants’ generalization in this domain. In the domain of stress assignment, we havesome preliminary evidence that multiple instances of a category (e.g., heavy syllable) areneeded for generalization. One question raised by the work on infants’ sensitivity to phonological form iswhether infants are in any sense biologically prepared to entertain certain categories ofsound experience, or whether any readily perceivable acoustic dimension can serve as a 188. 180 LouAnn Gerkenbasis for a category. Another way of framing the question is whether segments, segmentsequences, or stress assignment principles that are found among the world’s languagescan be learned more readily by human infants than other equally complex categoriesthat are not found in natural language. Researchers are just beginning to address thisquestion. At the present time, the answer appears to be that arbitrary patterns are learn-able (Chambers et al., 2003; Seidl & Buckley, 2005), although patterns that are char-acterizable in terms of disjunctions (e.g., words begin with /p/ or /s/) may suffer adisadvantage in generalization[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... (Saffran & Thiessen, 2003). Much more research isneeded on the sound stimulus properties required for generalization.Sensitivity to Syntactic FormAlthough the mantra of generative linguists over the past 50 years has been that syntaxis logically distinct from meaning (e.g., Chomsky, 1965), many diagnostics of syntacticstructure involve assessing meaning. For example, the fact that him in (2), below, cannotrefer to Bill is taken as evidence about the structural constraints on coreference. Surelyit makes little sense to assess learners’ sensitivity to such constraints before they canunderstand sentences like (2). (2) *Billi likes himi. Nevertheless, there are at least two aspects of syntactic sensitivity that can, in princi-ple, be assessed in the absence of sentence interpretation: word order and syntactic cate- gories. Although both of these components of syntax ultimately influence sentenceinterpretation, they can also be assessed to some extent on their own. For example,regardless of what meaning is intended, (3a) is not a possible sentence of English, becauseit violates English word order. Similarly, you may not know what zig, rif, or nug mean,but if you hear these words used in sentence (3b), you can feel confident that (3c) is agrammatical sentence. Researchers studying infants’ sensitivity to syntactic form havetaken advantage of these non-interpretational aspects of syntax to study early sensitivityto the orders of word-like units and to syntactic categories. (3) a. *Dog the cat the chased. b. The zigs were riffing the nugs. c. Look at those zigs rif.Sensitivity to the order of word-like unitsAs in the studies of sensitivity to phonological form, the first studies examining infants’sensitivity to the syntactic form involved the form of language the infant was alreadylearning. Shady, Gerken, and Jusczyk (1995) presented 10.5-month- olds with normalEnglish sentences as well as sentences in which determiners and nouns were reversed, 189. Acquiring Linguistic Structure 181resulting in phrases like kitten the. The stimuli were recorded using a speech synthesizerto avoid disruptions in prosody that are likely to occur when a human talker producesungrammatical sentences. Infants listened longer to the unmodified sentences, suggest-ing that they were able to tell the difference between the two types of stimuli. Similarstudies presented 10- to 12-month-old infants with normal English sentences versussentences in which a subset of grammatical morphemes was replaced by nonsense sylla-bles. Infants could discriminate the grammatical and ungrammatical stimuli (4a vs. 4b,below), but not stimuli in which nonsense words replaced content words (4a vs. 4c;Shady, 1996; Shafer, Shucard, Shucard, & Gerken, 1998). (4) a. There was once a little kitten who was born in a dark, cozy closet. b. There [ki] once [gu] little kitten who [ki] born in [gu] dark, cozy closet. c. There was once a little [mafIt] who was [tεk] in a dark, cozy closet. This pattern of results suggests that the information carried by grammatical mor-phemes was more salient to infants than particular content words, which they may ormay not have recognized. Santelmann and Jusczyk (1998) showed that 18-month-olds,but not 15- month-olds, are able to detect violations in dependencies between[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Englishmorphemes, such as auxiliary is and progressive suffi x -ing (e.g., Grandma is singing vs.Grandma can singing), only when the distance between the two morphemes was betweenone and three syllables. Although these studies indicate that infants are sensitive to aspects of their input thatmight serve as “cues” to an aspect of adult syntax, we cannot take such cue sensitivityto indicate that these infants have knowledge of English phrase structure. Rather, cuesensitivity merely indicates that infants have encoded frequently occurring patterns intheir native language. For example, in the Shady et al. (1995) study, many of the ungram-matical sentences contained two grammatical morphemes in sequence (e.g., a that). Suchsequences are virtually non-existent in English, and infants were probably respondingto this and similar aspects of the stimuli, as opposed to any tacit expectation for deter-miners to precede nouns. Because it is difficult to separate sensitivity to syntactic structure and frequencyof occurrence in the native language, researchers studying infants’ generalizationsover sentence-like stimuli have turned to familiarization studies like those discussedin the section on infants’ sensitivity to phonological form. In one such study, Gómezand Gerken (1999) presented 12-month-olds with a subset of strings produced byone of two finite state grammars. The two grammars began and ended in the sameCVC nonsense words, with the only difference being the string-internal sequences ofwords allowed. In one study, half of the infants were familiarized for about two minuteswith strings from Grammar 1 and half with strings from Grammar 2. For example,VOT PEL was a legal sequence in strings of Grammar 1, but not Grammar 2. Duringtest, both groups of infants heard new strings from the two grammars. Infants showeda significant preference for the new strings generated by their familiarization grammar.This study showed that infants learned about the sequential dependencies of thewords in their familiarization grammar and applied this knowledge to new stringsduring test. 190. 182 LouAnn Gerken Table 9.1 AAB familiarization stimuli used by Marcus et al. (1999) A B di je li we le leledi leleje leleli lelewe wi wiwidi wiwije wiwili wiwiwe ji jijidi jijije jijili jijiwe de dededi dedeje dedeli dedewe One important property of Grammar 1, and not Grammar 2, was that certain wordswere allowed to repeat in sequence. For example, VOT PEL PEL JIC was a legal stringin Grammar 1. By contrast, Grammar 2 contained strings in which the same wordoccurred in multiple string positions with other words intervening (e.g., PEL RUD JICVOT RUD). These repetitions and alternations might allow learners to recognize theabstract form of some of the strings in their familiarization language, even if the testitems contained new vocabulary. To test this possibility, Gómez and Gerken (1999)paired each word from the familiarization vocabulary with a new word in the testvocabulary (e.g., JED, FIM, TUP, DAK, SOG were matched with VOT, PEL, JIC,RUD, TAM, respectively). Thus, an infant who heard a string like JED-FIM-FIM-TUPin training might hear a string like VOT-PEL-PEL-JIC in test (both strings were gener-ated by Grammar 1). Again, infants showed a preference for strings that were consistentwith their familiarization grammar, suggesting that they[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... had discerned the pattern ofrepetitions and alternations of the two grammars (also see Gómez & Gerken, 1998). In a similar series of studies, Marcus, Vijayan, Rao, and Vishton (1999) exposed 7-month-olds to three-minute speech samples of strings with ABA (wi-di-wi and de-li-de)or ABB (wi-di-di and de-li-li) patterns. During test, infants heard strings with the samepattern they had heard during training as well as the other pattern, both instantiated innew vocabulary (e.g., ba-po-ba vs. ba-po-po). Infants trained on ABA stimuli preferredABB stimuli at test, while infants trained on ABB stimuli preferred ABA stimuli at test(i.e., a novelty preference). These results, coupled with those of Gómez and Gerken(1999), make clear that infants can generalize beyond specific word order based on pat-terns of repeating or alternating elements. A follow-up study using a subset of the stimuli used by Marcus et al. (1999) shedssome light on the conditions under which infants do and do not generalize beyond thespecifics of their input (Gerken, 2006). The stimuli from the AAB condition of theMarcus et al. study are shown in Table 9.1. If one considers all of the information inthe table, a succinct generalization is that all strings have an AAB form. The same istrue if one considers just the four stimuli on the diagonal. However, if one considersthe stimuli in the first column, all of the strings not only have an AAB form butalso end in the syllable di. Which generalization is correct? Recall, the observationthat a set of input data can give rise to multiple generalizations has been used as an 191. Acquiring Linguistic Structure 183argument that learners are innately constrained to make some generalizations and notothers (see above). To determine which generalization infants made, Gerken (2006) familiarized 9- month-olds with one of four sets of stimuli: AAB stimuli from the diagonal of Table 9.1,ABA stimuli from the diagonal (ledile, wijewi, jiliji, dewede), AAB stimuli from the firstcolumn of Table 9.1, and ABA stimuli from the first column (ledile, widiwi, jidiji, dedide).At test, infants heard new AAB and ABA strings. The rationale was if infants discernedeither an AAB or ABA pattern in the familiarization stimuli, they would be able to dis-criminate the new AAB and ABA test strings, replicating Marcus et al. (1999). In fact,only infants who were familiarized with stimuli from the diagonal discriminated the teststrings, suggesting that infants familiarized with the first column made the more local“contains di” generalization. This interpretation was confirmed in a second study, inwhich infants familiarized with the first column (either the AAB or ABA version) weretested on new strings in which the B element was the syllable di. In this study, infantswere able to discriminate AAB from ABA test stimuli. These studies suggest that thetype of form-based generalization learners make is very much dependent on the specificproperties of the input they encounter. Although we cannot yet determine how infantsselect one generalization out of a number of possibilities, the answer to that question willhelp us to compare nativist versus learning accounts of language development. The input required for infants to make a particular generalization has been exploredin another set of studies that focus on the[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... conditions under which infants discern longdistance dependency relations. Gómez (2002) familiarized 18-month-olds with an arti-ficial grammar of the form AXB and CXD, in which there is a dependency between theA and B elements and between the C and D elements. Importantly, she found that itwas only when the middle element was selected from a large pool (24) that infants coulddetect the relation between the first and third elements in the grammar. Gómez interpretsher result to mean that infants attempt to process the strings in terms of sequentialdependencies (A–X, X–B) until some point at which doing so becomes unfeasible. Thus,the processing resources required to encode stimuli in one manner versus another maybe one factor driving the particular generalizations that learners make.Sensitivity to syntactic categoriesResearchers have begun to examine, in addition to word order, infants’ sensitivity to thedistributional correlates of syntactic categories. The basic research strategy is to testinfants’ sensitivity to morpho-phonemic paradigms, as exemplified for Russian noungender in Table 9.2 (Gerken, Wilson, & Lewis, 2005). Seventeen-month-old infantswere familiarized for two minutes with the non-emboldened words in Table 9.2. Notethat if infants were able to detect that the case endings u and oj occurred on one setof words and ya and yem occurred on another set, they might be tacitly able to predictthe withheld emboldened words. During test, infants heard on alternate trials the gram-matical emboldened words and ungrammatical words created by combining masculinenouns with feminine case endings and vice versa. Infants were able to discriminate the 192. 184 LouAnn GerkenTable 9.2 Russian feminine and masculine nouns, each with two case endingsFeminine nounspolkoj rubashkoj ruchkoj vannoj knigoj korovojpolku rubashku ruchku vannu knigu korovuMasculine nounsuchitel’ya stroitel’ya zhitel’ya medved’ya korn’ya pisar’yauchitel’yem stroitel’yem zhitel’yem medved’yem korn’yem pisar’yemWords in bold were withheld during familiarization and comprised the grammatical test items. An apostropheafter a consonant indicates that the consonant is palatalized in Russian. Ungrammatical words were vannya,korovyem, medevedoj, pisaru.grammatical from ungrammatical items, suggesting that they had discerned the para-digm (Gerken et al., 2005). It is important to note that 12- month-old infants wereunable to discern the Russian gender paradigm shown in Table 9.2. However, infants atthat age demonstrate a potential precursor to the categorization ability shown by 17-month-olds (for details see Gómez & LaKusta, 2004). It is also important to note, however, that infants, like adults, were able to discriminategrammatical from ungrammatical items only when a subset of the words contained asecond cue to category membership. Note that a subset of the feminine words in Table9.2 end in k and a subset of the masculine words end in tel. Studies with adults andchildren tested in a paradigm completion format suggest that they too are unable todiscern the structure of a morpho-phonological paradigm unless morphological markersto categories are supplemented with semantics, phonology, or additional morphology(Braine, 1987; Frigo & McDonald, 1998; Mintz, 2002; Wilson,[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... 2000). Gerken et al.(2005) suggest that requiring multiple cues to syntactic categories protects learners fromovergeneralizing category structure. Other researchers have investigated infants’ sensitivity to morpho-phonological para-digms in their native language (Höhle, Weissenborn, Kiefer, Schulz, & Schmitz, 2004).Researchers familiarized 14- to 16-month-old German learners with two nonsense wordsin either a noun context (preceded by a determiner) or a verb context (preceded by apronoun). The infants then heard passages in which the new words were used as nounsor verbs. Infants who were familiarized with phrases in which the novel word was usedas a noun preferred passages in which it was used as a verb. These results suggest thatinfants track the morphological contexts that occur with particular nouns. When theyhear a new word in a noun context, they expect that the new word will also appear inother noun contexts. This expectation may be a sign of infants’ having formed proto-syntactic categories.Summary of infants’ sensitivity to syntactic formThe studies of infants’ sensitivity to syntactic form, like the studies examining sensiti-vity to phonological form, indicate that infants are skilled at detecting patterns in 193. Acquiring Linguistic Structure 185language and are able to generalize beyond the particular stimuli that they haveencountered when given appropriate evidence. The age at which sensitivity to pos-sible precursors of syntax appears varies considerably from 7 months (Marcus et al.,1999) to the middle of the second year (Gerken et al., 2005; Gómez, 2002; Höhleet al., 2004). As in the case of the studies on infants’ sensitivity to phonological form, we can askhow the data on infants’ sensitivity to syntactic form is related to what we know aboutsyntactic structures and categories in human language. The studies on word ordersuggest that infants are sensitive to the order of particular elements in a string, to pat-terns of repeating and alternating elements, and to correlations between non-adjacentitems. Although languages make some limited use of repeated morphemes (i.e., redupli-cation), repetition of the same word or morpheme is not typically viewed as central tomorpho-syntax. Similarly, long distance dependencies in natural languages occur acrossconstituents without a fi xed length (e.g., Arielle called her soccer-playing friend, Sara,up). The studies on infants’ sensitivity to morpho-phonological paradigms may betterreflect processes that occur in the acquisition of natural language syntax (e.g., Braine,1987). However, it is important to note that the syntactic categories derived via morpho-phonological cues are simply groups of words that appear in similar morpho-syntacticcontexts. That is, they are not labeled for the learner as noun, verb, etc. Therefore, intheories that hold labeled syntactic categories to be crucial (e.g., Baker, 2001), cate-gories created based on morpho- phonological cues alone may be of limited use (seePinker, 1984).What Early Sensitivity to Linguistic Form Tells Us aboutLanguage DevelopmentThroughout the chapter, hints about developmental sequence can be found in statementslike “17-month-olds, but not 12-month-olds[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... discriminated . . .”. The reader is cautionedthat almost none of these developmental differences is statistically reliable, therebymaking it very difficult in most of the research reported to determine a developmentaltimeline or developmental mechanisms. Nevertheless, we can attempt to construct arough timeline for the infant abilities discussed in this chapter. The studies on phonological form presented in this chapter suggest that 9-month- oldinfants are sensitive to the basic categories of phonology, including phonetic features andsyllable shapes, and to at least some of the principles by which these categories are com-bined. The studies on syntactic form suggest that infants are sensitive to the ordering ofword-like units by 7 months, and perhaps before. However, infants’ ability to track theinformation required to infer syntactic categories has not been shown in infants youngerthan 14 months. What do these studies, which demonstrate infants’ sensitivity to linguistic form inthe absence of meaning, tell us about language development? The ability to generalizebeyond our linguistic experience to produce and comprehend new utterances based on 194. 186 LouAnn Gerkenan arbitrary, multi-leveled, system has been taken as the great mystery of human lan-guage. The studies presented in this chapter demonstrate that infants have a remarkableability to keep track of the specifics of the form of their input and, importantly, to gen-eralize to new forms given sufficient evidence that generalization is warranted. Does thismean that the abilities documented in the studies presented here reflect the beginningsof language development? In considering the answer to that question, we must keep in mind that the ability togeneralize across fairly complex patterns is not the unique domain of human languagebut rather can be seen in a host of non-humans. For example, the types of generalizationby infants reported by Marcus et al. (1999) and Gerken (2006) can be seen in honeybees, pigeons, and cotton-top tamarins (Cumming & Berryman, 1961; Giurfa, Zhang,Jenett, Menzel, & Srinivasan, 2001; Hauser, Weiss, & Marcus, 2002). By contrast, otherhuman linguistic abilities may have no parallel in non-humans (e.g., Fitch & Hauser,2004; Hauser, Chomsky, & Fitch, 2002; Hauser, Newport, & Aslin, 2001). For example,humans, but not cotton-top tamarins, can learn a grammar of the form A nBn, whichgenerates strings like (AB, AABB, AAABBB, etc.). Because only humans have commu-nication systems with the power of human language, should we consider only abilitiesseen in humans and not other animals when we contemplate the mechanisms of languagedevelopment? If so, many of the studies reported here will ultimately be dismissed asirrelevant to language development, although they may inform us about human learningmore generally. Alternatively, we can view the process of language development as one in which learn-ers must use their pattern detection and categorization skills to discern the patterns andcategories employed by human language. On this view, some of these skills may well beshared by other species. A similar argument has been made about categorical perceptionfor speech sounds, which can be seen in species other than humans (e.g., Kuhl & Miller,1975). It[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... is now generally accepted that human speech perception has taken advantageof a general auditory property also found in other animals (Aslin, Pisoni, & Jusczyk,1983). One barrier to the view that language develops from the application of general,but powerful, learning mechanisms to linguistic data is the argument of “poverty of thestimulus,” which states that children are not exposed to linguistic structures of certaintypes that nevertheless appear to be part of their early knowledge of language (e.g.,Chomsky, 1980). If relevant input from which patterns and categories can be detecteddoes not exist, even the most computationally skilled learner cannot acquire a language.Although there is growing skepticism about the degree to which the input is trulyimpoverished (e.g., Elman, 2003; Lewis & Elman, 2001; Pullum & Scholz, 2002), muchmore work needs to be done to determine if there are indeed critical gaps in the infants’experience. In the mean time, however, explorations of infants’ sensitivity to linguistic formprovide us with a potentially important view of the infant’s world. They have the poten-tial to inform us about what abstract structures are relatively easy and difficult to detect.Further, as infant studies come systematically to examine sensitivity to forms like thosefound in languages of the world, they have the potential to change how we view humanlanguage and its development. 195. Acquiring Linguistic Structure 187ReferencesAslin, R. N., Pisoni, D. B., & Jusczyk, P. W. (1983). Auditory development and speech percep- tion in infancy. In M. M. Haith & J. J. Campos (Eds.), Handbook of child psychology: Infant development. New York: Wiley.Baker, M. C. (2001). The atoms of language. New York: Basic Books.Best, C. T. (1995). Learning to perceive the sound pattern of English. In C. Rovee-Collier & L. Lipsitt (Eds.), Advances in infancy research (pp. 217–304). Norwood, NJ: Ablex Publishing Co.Braine, M. D. S. (1987). What is learned in acquiring word classes – a step toward an acquisition theory. In B. MacWhinney (Ed.), Mechanisms of language acquisition (pp. 65–87). Hillsdale, NJ: Lawrence Erlbaum Associates.Chambers, K. E., Onishi, K. H., & Fisher, C. L. (2003). Infants learn phonotactic regularities from brief auditory experience. Cognition, 87, B69–B77.Chomsky, N. (1965). Aspects of the theory of syntax. Cambridge, MA: MIT Press.Chomsky, N. (1980). The linguistic approach. In M. Piattelli- Palmarini (Ed.), Language and learning (pp. 109–116). Cambridge, MA: Harvard University Press.Chomsky, N. (1981). Lectures on government and binding. Dordrecht: Foris.Cumming, W. W., & Berryman, R. (1961). Some data on matching behavior in the pigeon. Journal of the Experimental Analysis of Behavior, 4, 281–284.Cutler, A., & Carter, D. (1987). The predominance of strong initial syllables in the English vocabulary. Computer Speech and Language, 2, 133–142.Echols, C., Crowhurst, M., & Childers, J. B. (1997). The perception of rhythmic units in speech by infants and adults. Journal of Memory and Language, 36, 202–225.Eimas, P., Siqueland, E., Jusczyk, P. W., & Vigorrito, K. (1971). Speech perception in infants. Science, 171, 303–306.Elman, J. (2003). Generalization from sparse input. In Proceedings of the 38th Annual Meeting of the Chicago Linguistic Society.[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Chicago: University of Chicago Press.Fitch, W. T., & Hauser, M. D. (2004). Computational constraints on syntactic processing in a nonhuman primate. Science, 303, 377–380.Frigo, L., & McDonald, J. (1998). Properties of phonological markers that affect the acquisition of gender-like subclasses. Journal of Memory and Language, 39, 218–245.Gerken, L. A. (2004). Nine- month-olds extract structural principles required for natural lan- guage. Cognition, 93, B89–B96.Gerken, L. A. (2006). Decisions, decisions: Infant language learning when multiple generaliza- tions are possible. Cognition, 98, B67–B74.Gerken, L. A., Wilson, R., & Lewis, W. (2005). 17-month-olds can use distributional cues to form syntactic categories. Journal of Child Language, 32, 249–268.Gerken, L. A., & Zamuner, T. (in press). Exploring the basis for generalization in language acquisition. In J. Cole & J. Hualde (Eds.), LabPhon IX: Change in phonology. The Hague: Mouton de Gruyter.Giurfa, M., Zhang, S. W., Jenett, A., Menzel, R., & Srinivasan, M. V. (2001). The concepts of “sameness” and “difference” in an insect. Nature, 410, 930–933.Gómez, R. L. (2002). Variability and detection of invariant structure. Psychological Science, 13, 431–436.Gómez, R. L., & Gerken, L. A. (1998, April). Determining the basis of abstraction in arti- ficial language acquisition. Paper presented at the International Society on Infant Studies, Atlanta, GA. 196. 188 LouAnn GerkenGómez, R. L., & Gerken, L. A. (1999). Artificial grammar learning by 1-year-olds leads to specific and abstract knowledge. Cognition, 70, 109–135.Gómez, R. L., & Gerken, L. A. (2000). Infant artificial language learning and language acquisi- tion. Trends in Cognitive Sciences, 4, 178–186.Gómez, R. L., & LaKusta, L. (2004). A first step in form-based category abstraction by 12- month-old infants. Developmental Science, 7, 567–580.Guenther, F. H., & Gjaja, M. N. (1996). The perceptual magnet effect as an emergent property of neural map formation. Journal of the Acoustical Society of America, 100, 1111–1121.Guest, D. J., Dell, G. S., & Cole, J. S. (2000). Violable constraints in language production: Testing the transitivity assumption of Optimal Theory. Journal of Memory and Language, 42, 272–299.Hauser, M. D., Chomsky, N., & Fitch, T. (2002). The faculty of language: what is it, who has it, and how did it evolve? Science, 298, 1569– 1579.Hauser, M. D., Newport, E. L., & Aslin, R. N. (2001). Segmentation of the speech stream in a non-human primate: Statistical learning in cotton-top tamarins. Cognition, 78, B53–B64.Hauser, M. D., Weiss, D., & Marcus, G. F. (2002). Rule learning by cotton-top tamarins. Cognition, 86, B15–B22.Hirsh- Pasek, K., Kemler Nelson, D., Jusczyk, P. W., Wright Cassidy, K., Druss, B., & Kennedy, L. (1987). Clauses are perceptual units for prelinguistic infants. Cognition, 26, 269–286.Hogg, R., & McCully, C. B. (1987). Metrical phonology. Cambridge: Cambridge University Press.Höhle, B., Weissenborn, J., Kiefer, D., Schulz, A., & Schmitz, M. (2004). Functional elements in infants’ speech processing: The role of determiners in segmentation and categorization of lexical elements. Infancy, 5, 341–353.Hyams, N. (1986). Language acquisition and the theory of parameters. Dordrecht:[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Reidel.Jusczyk, P. W. (1985). On characterizing the development of speech perception. In J. Mehler & R. Fox (Eds.), Neonate cognition: Beyond the blooming buzzing confusion. Hillsdale, NJ: Lawrence Erlbaum Associates.Jusczyk, P. W., & Aslin, R. N. (1995). Infants’ detection of the sound patterns of words in fluent speech. Cognitive Psychology, 29, 1– 23.Jusczyk, P. W., Cutler, A., & Redanz, N. (1993). Infants’ sensitivity to predominant word stress patterns in English. Child Development, 64, 675– 687.Jusczyk, P. W., Friederici, A. D., Wessels, J. M., Svenkerud, V. Y., & Jusczyk, A. M. (1993). Infants’ sensitivity to the sound patterns of native language words. Journal of Memory and Language, 32, 402–420.Jusczyk, P. W., Luce, P. A., & Charles-Luce, J. (1994). Infants’ sensitivity to phonotactic patterns in the native language. Journal of Memory and Language, 33, 630– 645.Kuhl, P. K., & Miller, J. D. (1975). Speech perception in the chinchilla: Voiced–voiceless distinc- tion in alveolar plosive consonants. Science, 190, 69–72.Kuhl, P. K., Williams, K. A., Lacerda, F., Stevens, K. N., & Lindblom, B. (1992). Linguistic experience alters phonetic perception in infants by 6 months of age. Science, 255, 606– 608.Lewis, J. D., & Elman, J. L. (2001). A connectionist investigation of linguistic arguments from the poverty of the stimulus: Learning the unlearnable. In J. D. Moore & K. Stenning (Eds.), Proceedings of the Twenty-Third Annual Conference of the Cognitive Science Society (pp. 552– 557). Mahwah, NJ: Erlbaum.MacKain, C. (1982). Assessing the role of experience in infant speech discrimination. Journal of Child Language, 9, 527–542. 197. Acquiring Linguistic Structure 189Marcus, G. F., Vijayan, S., Rao, S. B., & Vishton, P. M. (1999). Rule learning by seven-month- old infants. Science, 283, 77–80.Maye, J., & Gerken, L. A. (2001). Learning phonemes: How far can the input take us? In H.-J. Do, L. Domínguez, & A. Johansen (Eds.), Proceedings of the 25th Annual Boston University Conference on Language Development (pp. 480–490). Somerville, MA: Cascadilla Press.Maye, J., & Weiss, D. J. (2003). Statistical cues facilitate infants’ discrimination of difficult phonetic contrasts. In B. Beachley, A. Brown, & F. Conlin (Eds.), Proceedings of the 27th Annual Boston University Conference on Language Development (pp. 508–518). Somerville, MA: Cascadilla Press.Maye, J., Werker, J. F., & Gerken, L. A. (2002). Infant sensitivity to distributional information can affect phonetic discrimination. Cognition, 82, B101–B111.Mintz, T. (2002). Category induction from distributional cues in an artificial language. Memory and Cognition, 30, 678–686.Naigles, L. R. (2002). Form is easy, meaning is hard: Resolving a paradox in early child language. Cognition, 86, 157–199.Pinker, S. (1984). Language learnability and language development. Cambridge, MA: Harvard University Press.Polka, L., & Werker, J. F. (1994). Developmental changes in perception of nonnative vowel contrasts. Journal of Experimental Psychology: Human Perception and Performance, 20, 421– 435.Pullum, G. K., & Scholz, B. C. (2002). Empirical assessment of stimulus poverty arguments. Linguistic Review, 19, 9–50.Saffran, J. R., & Thiessen, E. D. (2003). Pattern induction by infant language learners. Devel- opmental[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Psychology, 39, 484–494.Santelmann, L. M., & Jusczyk, P. W. (1998). Sensitivity to discontinuous dependencies in lan- guage learners: Evidence for limitations in processing space. Cognition, 69, 105–134.Sebastián Gallés, N., & Bosch, L. (2002). The building of phonotactic knowledge in bilinguals: The role of early exposure. Perception and Psychophysics, 28, 974– 989.Seidl, A., & Buckley, E. (2005). On the learning of arbitrary phonological rules. Language Learn- ing and Development, 3–4, 289–316.Shady, M. E. (1996). Infants’ sensitivity to function morphemes. Unpublished PhD dissertation, State University of New York at Buffalo, Buffalo, NY.Shady, M. E., Gerken, L. A., & Jusczyk, P. W. (1995). Some evidence of sensitivity to prosody and word order in ten-month-olds. In D. MacLaughlin & S. McEwan (Eds.), Proceedings of the 19th Boston University Conference on Language Development: Vol. 2. Somerville, MA: Cas- cadilla Press.Shafer, V. L., Shucard, D. W., Shucard, J. L., & Gerken, L. A. (1998). An electrophysiological study of infants’ sensitivity to the sound patterns of English speech. Journal of Speech, Language, and Hearing Research, 41, 874–886.Stager, C. L., & Werker, J. F. (1997). Infants listen for more phonetic detail in speech perception than in word-learning tasks. Nature, 388, 381–382.Thiessen, E. D., & Saffran, J. R. (2003). When cues collide: Use of stress and statistical cues to word boundaries by 7- to 9-month-old infants. Developmental Psychology, 39, 706– 716.Turk, A., Jusczyk, P. W., & Gerken, L. A. (1995). Infants’ sensitivity to syllable weight as a determinant of English stress. Language and Speech, 38, 143–158.Werker, J. F., Fennell, C., Corcoran, K., & Stager, C. L. (2002). Infants’ ability to learn phoneti- cally similar words: Effects of age and vocabulary size. Infancy, 3, 1–30. 198. 190 LouAnn GerkenWerker, J. F., & Pegg, J. E. (1992). Infant speech perception and phonological acquisition. In C. A. Ferguson, L. Menn, & C. Stoel-Gammon (Eds.), Phonological development: Models, research, implications (pp. 285–311). Timonium, MD: York Press.Werker, J. F., & Tees, R. C. (1984). Cross-language speech perception: Evidence for perceptual reorganization during the first year of life. Infant Behavior and Development, 7, 49–63.Wilson, R. (2000). Category learning in second language acquisition: What artificial grammars can tell us. Unpublished Masters thesis, University of Arizona, Tucson, AZ. 199. 10Cognitive Processes in Early Word LearningDiane Poulin-Dubois and Susan A. GrahamIn this chapter, we examine the challenging issue of how children learn their first wordsand become master word learners before they can tie their shoes. By the end of the firstyear, infants have acquired sophisticated speech perception skills and have started toutter conventional words. They are also budding scientists, having developed some primi-tive “theories” about human psychology and object physics. The next challenge facingthe child is to bridge the gap between words and concepts in order to establish whatwords mean. There are many questions surrounding this important milestone:What words do children learn first and why are some[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... words easier to learn than others?What do children know about the meanings of words? What are the cognitive mecha-nisms involved in learning words? We address these questions by reviewing research onearly lexical development and cognitive development in order to uncover the tools infantshave at their disposal to acquire a lexicon. We review research on how cognitive skills,both domain-general and domain-specific (those skills tailored to process only certainkinds of information), are instrumental in helping children solve the formidable induc-tion problem posed by word learning, including research on object and event representa-tions and categories, as well as precursors to a theory of mind.A Brief Historical ReviewThe notion that the development of cognitive abilities provides a foundation for theacquisition of language has been unanimously shared by developmentalists interested inearly semantic development for many years (e.g., Clark, 1983; Dromi, 1993). One of themost influential proponents of this view was Jean Piaget. An important contributionof Piaget’s work was to demonstrate first, that infants develop concepts and problem-solving skills during the first 18 to 24 months of life (the sensorimotor period). Piaget 200. 192 Diane Poulin-Dubois and Susan A. Grahamfurther demonstrated that these cognitive abilities provide the foundations for the emer-gence of symbolic representations, as expressed in language, pretend play, and othersymbols. Piaget’s perspective on the relationship between language and cognitioncontrasted sharply with the nativist view on language that also emerged at that time(Piatelli-Palmarini, 1980). Despite the fact that Piaget’s theory did not address directly the task of word learning,his theory contributed to the emergence of a strong cognitive hypothesis. According tothis proposal, conceptual notions serve as the prerequisites for acquiring the linguisticforms that encode them (e.g., MacNamara, 1972). This proposal was soon followed, inthe 1970s and 1980s, by a flurry of studies on the relations between linguistic and cogni-tive developments in infancy (e.g., Bates, Benigni, Bretherton, Camaioni, & Volterra,1979; Gopnik, 1984). Many of these studies examined the relationship between infants’general sensorimotor development and general measures of early language such as meanlength of utterance, age of acquisition of first words, or vocabulary size. In general, theresults of these studies indicated a weak correspondence between language and cognition,although the predictive value of conceptual skills increased significantly when specificlinguistic and conceptual milestones were considered. This finding led some researchersto propose that language and cognition are interrelated only when specific tasks areconsidered, a proposal that has been referred to as the specificity hypothesis (Gopnik &Meltzoff, 1986). An often-cited piece of evidence for this hypothesis concerns the linkbetween the naming explosion typically observed around 18 months of age and a newmilestone in the ability to categorize objects (Gopnik & Meltzoff, 1987; Poulin-Dubois,Graham, & Sippola, 1995, but see Gershkoff- Stowe, Thal, Smith, & Namy, 1997). Over the last two decades, the issue of how language acquisition builds on cognitivedevelopment has taken new[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... directions. One direction has been to specify the constraintsthat children honor when learning a new word (e.g., Markman, 1994; Woodward &Markman, 1998). The purely linguistic nature of these constraints is a topic of contro-versy (Bloom, 2000). Another line of research has examined the cognitive prerequisitesfor acquiring specific word forms. That is, it has been proposed that the acquisition ofthe meanings of nouns will be much more dependent on cognitive than linguistic factorsrelative to the acquisition of other words (Gentner & Boroditsky, 2001). A third researcharea concerns the cognitive factors that influence how infants generalize words to newreferents. In particular, research on infant categorization has provided new informationregarding the basis for children’s word usage in the second year of life. Finally, the roleof social cognitive abilities in assisting infants in some stages of lexical developmenthas received a great deal of empirical attention (e.g., Baldwin & Meyer, this volume;Tomasello, 2001).Types of Words in Infants’ LexiconWhile infants’ vocabularies contain a variety of words, the early vocabulary of youngword learners has often been characterized as biased toward nouns. Nouns form themajority of children’s early receptive and productive vocabulary and are typically acquired 201. Cognitive Processes in Early Word Learning 193earlier than other word classes in many languages (Bates et al., 1994; Bloom, 1998;Fenson et al., 1994). As infants acquire more vocabulary, the gap between nouns andother word forms begins to close. The main theoretical argument for the early dominance of nouns emphasizes con-ceptual factors, positing that it is easier to acquire labels for objects than labels for verbs.That is, it is proposed that nouns are easier to learn because they refer to perceptuallydistinct and coherent units that are stable and consistent across time and context(Gentner, 1982). By contrast, the task involved in learning a label for an action is acognitively more complex one, as the child needs to abstract the constant elements acrossa variety of contexts labeled by the verb, and understand the particular relationshipbetween subject and object (Gentner, 1981; MacNamara, 1972). In a recent paper,Gentner and Boroditsky (2001) have developed and expanded upon Gentner’s (1982)original position by proposing the “division of dominance” hypothesis, which posits thatwords vary along a continuum of cognitive versus linguistic dominance. On the cognitiveend of the continuum lie words that refer to perceptually “individuated” items (i.e.,concrete nouns). Words that cannot “exist independently of language” are at the linguis-tic end of the continuum (i.e., determiners and conjunctions). Verbs lie somewhere inthe middle of this continuum, as languages vary in the way they choose to lexicalize andpackage the same event. Consequently, it is argued that verbs are acquired later, andpreviously learned lexical items, such as noun–object pairs, influence verb learning(Gentner & Boroditsky, 2001). Despite the empirical evidence and theoretical support for dominance of nouns inearly vocabulary, it has been argued that the noun bias might simply be an artifact ofthe linguistic structure of English, rather than a universal in infants’ vocabularies[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... (Choi& Gopnik, 1995; Tardif, Gelman, & Xu, 1999). Proponents of this view have arguedthat infants’ early lexicon reflects the linguistic input to which they are exposed, callingattention to the fact that the structural properties of languages differ in their emphasison nouns. For example, in English, names for objects are most likely to be the loudestelement of a sentence and they are often found in sentence final position (Tardif, Shatz,& Naigles, 1997). These characteristics of English may make nouns the most salientpart of the sentence, making it easier for children to attend to them (Slobin, 1973). Thisis in sharp contrast to SOV (Subject–Object–Verb) languages like Korean and Japanese.Despite this debate, it remains that the “noun bias” has been reported in a wide rangeof languages other than English, including French, Italian, Spanish, Hebrew, Dutch,Korean, and Mandarin (Au, Dapretto, & Song, 1994; Bornstein et al., 2004; Caselliet al., 1995; Dromi, 1987; Gentner, 1982; Jackson-Maldonado, Thal, Marchman, Bates,& Gutierrez-Clellen, 1993; Poulin-Dubois et al., 1995; Tardif et al., 1999). In sum, the universal presence of many different word types in the early lexicon sug-gests that children create meaning to map onto different word forms at the very firststages of lexical development. To do so, they have to draw on the repertoire of conceptualknowledge that they have built over their first year of life, such as knowledge aboutobjects, actions, and properties and how they map to different linguistic categories likenouns, verbs, and adjectives (Clark, 1993). In the next sections, we review the domain-general as well as domain- specific cognitive processes that are utilized by children tolearn each of these word forms. 202. 194 Diane Poulin-Dubois and Susan A. GrahamDomain-General Processes in Learning WordsThere is a large literature that indicates that domain-general cognitive mechanisms (e.g.,learning processes, memory, attention, social-cognitive skills, etc.) facilitate all aspectsof language learning, including speech segmentation, phonetic and grammatical catego-rization, and word learning (see Baldwin & Meyer, this volume; Saffran & Thiessen,this volume). With regard to word learning, researchers have proposed that many wordlearning skills are domain-general processes that are recruited in a wide range of learningcontexts (e.g., Bloom, 2000; Samuelson & Smith, 2000). For example, mechanisms oflearning, memory, and retrieval, which are well known in the adult verbal learning lit-erature, also apply to the early stages of word learning. This is well demonstrated byoverextension errors which occur when the child knows the correct word but accessesthe wrong word by mistake (e.g., dog for horse). These errors appear to reflect a difficultyin retrieving a known word from the lexicon and are more frequent at the time of rapidvocabulary growth (Gershkoff-Stowe, 2002; Huttenlocher, 1974). Sensitivity to atten-tional cues, such as gaze direction, is another case of a domain-general mechanism whichcan be instrumental in establishing the reference of a new word and in distinguishingbetween accidental and intentional actions (Baldwin, 1993; Woodward, 2004). In fact,some researchers have argued that specialized developmental mechanisms for[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... learningwords are the product of language development, with general and dumb processes ini-tially driving word generalizations (Smith, 2001). In this chapter, we presume that chil-dren do bring a powerful set of domain-general learning processes to the task of learningwords, including memory capabilities, categorization skills, selective attention, and asensitivity to the intentions of others. In the sections that follow, we review how specificcognitive accomplishments assist infants in acquiring specific types of words.Cognitive Processes in Learning NounsAs described earlier, words for objects, or nouns, occupy a privileged position in the earlylexicon. Not surprisingly, a great deal of research has been devoted to understanding theprocesses that underlie the acquisition of nouns (see Bloom, 2000; Hall & Waxman,2004, for reviews). In this section, we review the cognitive processes that underlie objectword learning, focusing on infants’ understanding of objects and object categories.Object representationsOne of the initial steps in learning an object word is to identify the referent of a wordamongst the many possible candidates present in any given situation. The word learnermust then make a link between that word and the object. In order to make a reliablemapping between a word and an object, infants must build a representation of the objectthat is the referent of the new word. This process of building an object representation 203. Cognitive Processes in Early Word Learning 195rests upon infants’ ability to view an object as a solid body that continues to exist whenoccluded and that maintains its identity over time. In this section, we review evidenceindicating that well before they reach their first birthdays, infants have developed animpressive understanding of objects. In order to determine which object is the intended referent of the new word, infantsmust be able to identify the often numerous objects present in the scene. Thus, theyfirst must parse the surfaces present into distinct entities, a task referred to as objectsegregation. For example, they must appreciate that the dog partially hidden behind atree is one dog, rather than multiple objects. Second, they must be able to determinethe number of distinct entities present in any given scene and must track the identityof these objects through space and time, a task referred to as object individuation.For example, they must recognize that the dog they saw behind the tree is the samedog that is now running across the field. Research has demonstrated that already before12 months of age infants are highly skilled at both of these tasks. That is, they canuse a variety of perceptual cues for object segregation, including motion of visible sur- faces and edge alignment (e.g., Johnson, 2004). Similarly, infants can utilize severaltypes of information when individuating objects, including spatiotemporal information,featural information, and kind information (see Wilcox, Schweinle, & Chapa, 2003;Xu, 2003). In addition to the ability to segregate and individuate objects, infants need to appreci-ate some fundamental aspects of objects, such as object permanence. Seminal researchby Baillargeon and her colleagues with procedures based on visual attention has dem-onstrated that infants as young as 2.5 months of age can[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... appreciate that objects continueto exist when hidden (e.g., Baillargeon, 2004). Furthermore, infants can represent thelocations of hidden objects and they view objects as cohesive, three-dimensional bodiesthat trace continuous paths in space and time (Spelke & Van de Walle, 1993). Finally,infants appear to have a basic understanding of physical properties of objects, includinggravity, and causality (Cohen & Oakes, 1993; Leslie & Keeble, 1987). In sum, researchindicates that infants in the first year of life can segregate, individuate, and representobjects.Object categoriesAnother important step in the word learning process involves generalizing a newly learntword to other appropriate instances of the referent when encountered in the future. Toaccomplish this task, infants require some understanding of the category or concept thatis marked by that word. How early in their word learning career infants can do so is thesubject of debate. Some researchers believe that some early word learners can learn onlycontext-bound words whereas others argue that infants can extend linguistic knowledgeto new contexts as soon as they comprehend their first words around 9 months (Barrett,1995; Schafer, 2005). In this section, first, we review evidence indicating that preverbalinfants have well-developed categorization abilities; second, we review research describ-ing the developmental path that infants follow when attaching words to categories, andthird, we describe the nature of early lexical object categories. 204. 196 Diane Poulin-Dubois and Susan A. Graham Well before they utter their first word, infants are already highly skilled at objectcategorization. For a young child to engage in categorization, he or she must perceiveentities as alike, thereby employing some type of similarity rule to group them together.Studies have demonstrated that by 2 to 3 months of age, infants can learn global categoryrepresentations, such as a category of furniture that includes novel furniture items butexcludes mammals. By 3 to 4 months of age, infants can also form basic-level categoryrepresentations, such as a category that includes novel cats but excludes birds, dogs, andhorses. By 6 to 7 months of age, infants can form more specific, subordinate-like catego- ries, such as a category of tabby cats that excluded Siamese cats (see Quinn, 2004, fora review). Of course, these early perceptual categories demonstrate only that infants areadept at extracting common properties across a wide range of objects. Infants’ objectcategories undergo further development during the second year of life with basic-levelcategories developing after global categories as shown by more “conservative” categoriza-tion tasks, such as object examination and sequential touching (e.g., Mandler, 2003). Infants not only have well-developed categorization abilities, but they can also usethe categories they form to guide their inductive inferences. Inductive inferences typicallyinvolve the following line of reasoning: first, observing that X has the property Y (e.g.,a robin can fly); second, deciding that X and Z are the same kind of thing (e.g., a robinand a sparrow are both birds); and third, inferring that Z also has the property Y (e.g.,therefore a sparrow can fly). Following from Susan Gelman’s seminal[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... research on pre-schoolers’ inductive abilities (see Diesendruck, this volume), studies have demonstratedthat infants as young as 9 months of age possess basic inductive reasoning abilities. Forexample, Baldwin, Markman, and Melartin (1993) found that 9- to 16-month-oldinfants generalized nonobvious object properties to objects that were perceptually similarto a target object (see also Graham, Kilbreath, & Welder, 2004; Welder & Graham,2001). Recent research has focused on the nature of the categories that guide infants’inductive generalizations. For example, infants as young as 9 months can generalize“animal” properties (e.g., drinking) to other animals and “vehicle” properties (e.g., beingkeyed) to other vehicles (Mandler & McDonough, 1996, 1998). Furthermore, infantswill not cross category boundaries and imitate actions appropriate for animals on vehiclesand vice versa.Linking words to objectsAs infants cross the word learning threshold, they appear to begin with a broad expecta-tion linking words to commonalities among objects, rather than with a specific expecta-tion that words link to category-based commonalities (Waxman, 2004; Waxman &Lidz, 2006). That is, infants begin with the notion that words highlight many differenttypes of commonalities amongst objects including category-based commonalities (e.g.,dogs, cats), event-based commonalities (jumping, rolling), and property- based com-monalities (e.g., rough things, smooth things). In support of this proposal, Waxman andBooth (2003) found that 11-month-old infants construed the same set of objects (e.g.,four purple animals) as either members of the same category (e.g., animals) or as allpossessing a salient property (e.g., purple things). Moreover, infants linked novel words, 205. Cognitive Processes in Early Word Learning 197presented either as nouns or as adjectives, to either construal. Waxman and her colleagueshave argued that this initial broad link serves a number of critical functions includingfacilitating the formation of many different concepts and assisting the infant in discover-ing more specific mappings between specific types of words (e.g., nouns, adjectives) andspecific types of relations they mark (e.g., noun– object categories, adjectives–objectproperties). This broad expectation that words map to a variety of commonalities becomes moreprecise as infants establish their lexicon (Waxman, 2004; Waxman & Lidz, 2006). Thatis, around 13 to 14 months of age, infants begin to map novel nouns to category-basedcommonalities amongst objects. Waxman and Booth (2001) demonstrated that 14-month-olds map novel nouns to category-based commonalities amongst objects (e.g.,horses) and not to property-based commonalities (e.g., purple things). Interestingly,infants mapped novel adjectives to either property-based or category-based commonali-ties. Thus, their expectations for novel nouns are more precise than their expectationsfor novel adjectives at 14 months of age.Nature of Early Lexical CategoriesIn recent years, a great deal of empirical attention has been devoted to investigating thenature of infants’ categories with specific focus on the characteristics of the similarityrules that are used to guide categorization (see Rakison & Oakes, 2003, for a review).In this section, we focus only on a[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... particular type of lexical category, namely, categoriesthat are labeled by nouns, and we describe several conclusions regarding the nature ofthis particular type of lexical category during infancy. First, perceptual features, in particular shape similarity, can play an important rolein the organization of infants’ lexical categories about objects. Several studies havedemonstrated that infants tend to privilege shape information over other types of per-ceptual information such as color and size when extending novel words (e.g., Gershkoff-Stowe & Smith, 2004; Graham & Poulin-Dubois, 1999). Although there is littlecontroversy around the finding that infants’ lexical categories can be organized by shape,there is significant debate regarding why children attend to shape when extending novelwords. Some researchers suggest that children attend to shape similarity when categoriz-ing objects as their categories reflect primarily their attention to salient perceptual fea-tures (e.g., Jones & Smith, 1993). By contrast, other researchers suggest that childrenattend to shape information when categorizing objects because it serves as a perceptuallyavailable cue to the kind to which an object belongs (e.g., Bloom, 2000; Gelman &Diesendruck, 1999). In support of the latter position, research has demonstrated thatinfants expect objects that share a high degree of shape similarity also to share nonobvi-ous properties, suggesting that they appreciate that shape similarity is a reliable cue tocategory membership (e.g., Graham et al., 2004; Welder & Graham, 2001). Second, and importantly, infants are not limited to grouping objects together on thebasis of perceptual features when forming lexical categories of objects. For example,studies have demonstrated that when objects are named, 14- month-old infants will 206. 198 Diane Poulin-Dubois and Susan A. Grahamcategorize objects that share less perceptually obvious features such as function (e.g.,Booth & Waxman, 2002; Welder & Graham, 2006). Similarly, Waxman and her col- leagues have demonstrated that naming objects with the same label will aid infants informing superordinate categories of objects that may share minimal perceptual similarity(Waxman & Markow, 1995). Finally, when 13- and 18- month-old infants are providedwith information about category membership in the form of shared object names, theyassume that two perceptually dissimilar objects belong to the same category and thereforeshare a nonobvious property (Graham et al., 2004; Welder & Graham, 2001). Thus,when forming a lexical category, infants will treat object names as marking categorymembership and will use that information to group together objects that may not shareperceptual similarity. Importantly, sound tones and emotional expressions (e.g., “Ah”)do not appear to have the same facilitative effect on categorization as words (Balaban &Waxman, 1997; Xu, 2002). In support of the integration of both perceptual and categori-cal knowledge in early word meaning, toddlers appear to make many overextensionerrors in production (e.g., referring to a pomegranate as an apple) that are not alwayserrors of use but also reflect underlying broad word meanings based on shape and taxo-nomic relatedness (Gelman, Croft, Fu, Clausner, &[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Gottfried, 1998). Finally, recent research has demonstrated that the associations children make betweena word and its referent in early word learning may be explained by a prototype framework(Barrett, 1995). For example, around 12 months of age, infants may first map a wordto a typical member of a category (e.g., a golden retriever) and not to an atypical memberof the category (e.g., a pug). By 18 months, however, infants broaden their exten-sions to include less typical category members (Meints, Plunkett, & Harris, 1999).Similarly, infants as young as 18 months are flexible in their object word extension inthat they will consider intact (e.g., a dog) and incomplete objects (e.g., a dog withouta tail and legs) as equally acceptable referents for familiar labels (Poulin-Dubois &Sissons, 2002). In sum, there is a great deal of evidence indicating that infants are highly skilled atbuilding object representations and forming object categories. These abilities set the stagefor the rapid acquisition of object words that takes place during infancy.Cognitive Processes in Verb LearningAlthough some verbs appear early in the lexicon, verbs seem to be universally harder tolearn than nouns (Bornstein et al., 2004). The problem in verb learning seems to beabout mapping a verb onto an action or event rather than about learning the underlyingconcepts that verbs encode (Maguire, Hirsch-Pasek, & Golinkoff, 2006). To learn averb, children have to determine which aspect of an ongoing event is being referred to,or to solve what Tomasello (1995) has called the “packaging problem.” One source ofthe difficulty naïve learners encounter in solving the verb packaging problem is that verbscan be interpreted in terms of numerous semantic elements such as manner of motion(e.g., walk vs. run), direction relative to the speaker (e.g., come vs. go), the instrumentinvolved (spoon vs. pedal), or by the result achieved (e.g., fill vs. empty) to name but a 207. Cognitive Processes in Early Word Learning 199few (Gentner, 1982; Talmy, 1985). In addition to learning which aspects of a verb’s refer-ent become conflated with meaning, children must also interpret actors’ behavioralintentions and speakers’ semantic intentions, as well as relying on their own semanticand syntactic understanding to overcome the referential obstacles to acquiring verbmeaning (Forbes & Farrar, 1995). In this section, we discuss how early action verbs arerepresented as well as the recent studies on infants’ understanding of human action andobject motion.Infants’ representations of verbsThe few studies that have focused on young children’s initial representation of verbmeaning indicate that verbs are initially narrowly defined and context-specific (Forbes& Farrar, 1993; Forbes & Poulin-Dubois, 1997). Children initially represent action verbmeaning in terms of event appearance, that is, the overall configuration of an action. Bythe end of the third year, children’s representation of verb meaning includes proportion-ally more defining elements of semantic meaning, including the actor’s intentions(Poulin-Dubois & Forbes, 2002, 2006), speaker’s semantic intentions (e.g., Forbes,Ashley, & Martin, 2003), as well as syntactic form and function relationships (e.g.,Naigles & Terrazas, 1998; Slobin, 2001). Recent experimental research has demonstrated that children seem to take more[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... timeto learn labels for actions than for objects. For example, when 2.5-year- olds are taughtnovel labels for novel objects and novel actions, children learn the noun–object associa-tion more easily than the verb–action association (Childers & Tomasello, 2002). Asimilar advantage of nouns over verbs is observed in studies that have examined howwell children generalize newly learned nouns and verbs to new instances. Both English-and Japanese- speaking preschoolers fail to extend a new verb to other similar actionswhen the object that is acted on is changed. By contrast, they extend nouns on the basisof similarity of objects independent of the action in which the object is used (Imai,Haryu, & Okada, 2005; Kersten & Smith, 2002). In a recent direct test of the word–object versus word–action association, 18- to 20-month-old infants were first familiar-ized with novel labels (e.g., fep) for computer- animated motion events and then showntest events in which either the motion or the object were switched in the presence of theoriginal label (Katerelos, Poulin-Dubois, & Oshima-Takane, submitted). The resultsindicated that French-, English-, and Japanese-speaking infants mapped the novel labelpreferentially to the object. Because the linguistic properties of Japanese (e.g., argument-dropping, SOV language) would predict that verb learning would precede noun learning,these findings provide strong evidence for the universal noun advantage. Furthermore, young children tend to rely on general-purpose verbs at the early stagesof verb acquisition (e.g., go, get). These findings may be due to the fact that categoriesof actions are less coherent than categories of objects. That is, action categories differfrom object categories in that first, actions involve a relation between at least one agentand some activity, and second, action categories are not as well defined as object catego-ries. For example, it is not clear when the action of picking up an object begins. Doesit begin when the hand starts to move toward the object, when the object is grasped, or 208. 200 Diane Poulin-Dubois and Susan A. Grahamwhen it is lifted off the floor? Finally, the “glue” that holds exemplars of an action cate-gory together is less binding than in the case of object categories. For example, in thecase of transitive actions, there is great diversity in the objects involved in actions sharingthe same label, as illustrated by activities labeled “holding,” including a vase holdingflowers, a mother holding a child, and a barrette holding hair (Clark, 1993).Infants’ understanding of motion eventsWhile some of the conceptual foundations for verb learning do overlap with those fornoun learning, there is knowledge that will be uniquely recruited for verb learning. Inparticular, in addition to the ability to attend to, individuate, and categorize actions,infants also have to determine which elements of meanings are encoded in a verb tosuccessfully map words to actions (Golinkoff et al., 2002). Since most early verbs denote concrete events, the processing of object motion andhuman action is required in order to learn verbs efficiently. A large body of researchindicates that preverbal infants can discriminate and conceptualize motion events. First,motion attracts the attention of very young infants. For example, infants of 2 to 3 monthsof age prefer to look at moving[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... rather than stationary objects (Kellman & Banks, 1998;Slater, 1989). Infants as young as 3 months of age can extract motion commonalitieswhen presented with point-light displays of the pendular motion of animals walking inplace or the rotary motion of vehicles rolling in place (Arterberry & Bornstein, 2001,2002). During the first year, infants also show sensitivity to a wide range of motionproperties, such as trajectory and onset of motion. For instance, infants as young as 3months discriminate between a point-light display representing a walking person and anincoherent display (Fox & McDaniel, 1982). Movement is also instrumental in thedevelopment of infants’ understanding of objects’ properties (see Baillargeon, 2004, fora review) and is critical in the perception of object unity (Johnson & Aslin, 1995; Smith,Johnson, & Spelke, 2003). By 6 months of age, infants can discriminate subtle differ-ences between action events, such as causal versus noncausal launching events and con-tingent versus noncontingent motion (Bahrick, Gogate, & Ruiz, 2002; Cohen & Amsel,1998). Infants of this age also discriminate manner and path in a motion event (Casasola,Bhagwat, & Ferguson, 2006) and can categorize dynamic point-light displays of themotion patterns of animals and vehicles (Arterberry & Bornstein, 2002). Using geomet-ric figures as stimuli, studies have demonstrated that by the end of the first year, infantscan categorize events based on path across varying manners and based on manner acrossvarying paths (Pulverman, Hirsh- Pasek, Golinkoff, Pruden, & Salkind, 2006). Not only do infants make fine discriminations of motion patterns, but infants areadept at generalizing motion across a wide range of agents, paving the way for actionword extension (Poulin-Dubois & Vyncke, 2005). In one study, we tested 14- and 18-month- olds’ ability to associate animals and people with animate motions (e.g., climbingstairs, jumping over a block) and vehicles with inanimate motions (e.g., jumping acrossa gap, sliding along a U-shaped block). An experimenter modeled each action three timeswith an appropriate target exemplar with each demonstration accompanied by a specificvocalization (e.g., making a dog walk up the stairs while saying “Tum, tum, tum”). 209. Cognitive Processes in Early Word Learning 201Infants were then given two test exemplars to imitate the action (e.g., a horse and a car).Infants as young as 14 months generalized motion trajectory from one category exemplarto a member of the same object kind. Interestingly, in another experiment, infants werealso able to generalize the actions modeled with an animal to both a person and anotheranimal, showing a broad concept of agents. Another recent series of experiments usingsimilar stimuli with the infant-controlled habituation paradigm provides convergingevidence that the ability to generalize motion properties across objects emerges duringthe second year (Baker, Demke, & Poulin-Dubois, submitted). For example, whenhabituated to films featuring a dog jumping over a wall and a car bouncing off the wall,infants as young as 12 months dishabituated to incongruent test events (e.g., bus jumpingover the wall) but not to congruent events (e.g., cat jumping over the wall). In sum, by the beginning of the[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... second year infants are not only competent in dis-criminating human actions and object motion, but they also understand that many dif-ferent agents are capable of performing the same actions. This knowledge makes possiblethe extension of action verbs to many different agents at a very early stage in lexicaldevelopment, analogous to the way nouns are extended to object categories.Infants’ understanding of the intentional nature of actionsAlthough motion and action processing constitute the building blocks from whichinfants can learn the meaning of many concrete action verbs (e.g., run, jump), othercognitive abilities are required to establish the reference of other verbs, such as spill andpour. To learn these verbs, infants must be able to represent human actions not only asphysical motion through space but also as having an intentional structure. Although thelexicons of very young infants do not contain many instances of this type of verb, thereis evidence that children can learn verbs like spill and pour toward the end of the secondyear (Poulin- Dubois & Forbes, 2002, 2006). Furthermore, by the beginning of thesecond year, children have begun to produce mental terms, more specifically terms thatrefer to desires (e.g., want, like), with other mental verbs emerging around the thirdbirthday (Bartsch & Wellman, 1995; Shatz, Wellman, & Silber, 1983). A number of recent studies have examined when infants represent actions in terms ofintentional relations (see Buresh, Woodward, & Brune, 2006, for a review). By 5 or 6months of age, infants are attentive to grasping events and represent them as object-directed events (Woodward, 1998, 1999). Interestingly, the same object-grasping actionsare not interpreted as object- directed if an inanimate object is the agent or if the handonly touches the object (Woodward, 1999; Woodward & Somerville, 2000). A fewmonths later, infants can infer goals from gaze or pointing alone, without physicalcontact with the object (Sodian & Thoermer, 2004; Woodward, 2003). By 10 or 11months of age, infants can parse an ongoing stream of action by using the beginningand end of intention-in-actions as boundaries. For example, when pauses are inserted invideos of everyday events (e.g., cleaning the kitchen), infants show more interest whenthe pauses occur in the middle of an intentional action than if they are inserted afteror before the action (Baldwin, Baird, Saylor, & Clark, 2001). When asked to imitatethe actions of other people, infants are also sensitive to behavioral cues that mark the 210. 202 Diane Poulin-Dubois and Susan A. Grahamunderlying goals of people’s action. For example, after observing an actor who tried, butfailed, to complete an action, 18-month-olds (but not 12-month-olds) are as likely toproduce the target action as those who had seen a demonstration in which the actorachieved his goal (Bellagamba & Tomasello, 1999; Meltzoff, 1995). When infantsranging from 14 to 18 months of age are requested to imitate an actor’s actions on objects,they are more likely to imitate actions accompanied by behavioral cues that suggestintentional actions (i.e., actor said “There!” and looking at the object while completingthe action) than actions that appear accidental (i.e., actor said “Whoops!” while lookingaway as she completes the action) (Carpenter, Nagell, & Tomasello, 1998; Olineck &Poulin-Dubois,[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... 2005). Finally, it has recently been shown that in a toy request task,infants as young as 9 months react with more frustration if a person appears to refuseto hand a toy to the child than if she seems to fail to hand the toy because she is clumsy(Behne, Carpenter, Call, & Tomasello, 2005). Taken together, this research suggeststhat infants possess the necessary cognitive skills to represent actions in terms of theirintentional structures by the time that action words enter their receptive vocabulary. Although infants’ early vocabularies are composed mainly of nouns and verbs, othertypes of words are present as well. Moreover, these other words can account for a sub-stantial portion of the vocabulary by the end of the infancy period. Although there isrelatively little research on word meanings other than nouns and verbs, researchers havestarted to examine the cognitive foundations for these words. We review some of theavailable evidence in the next section, focusing on recent studies on spatial preposi-tions (see Waxman & Lidz, 2006, for a review of the processes underlying adjectiveacquisition).Learning Spatial WordsIn many languages, including English, Italian, Turkish, and Hebrew, the spatial preposi-tions “in,” “on,” and “under” are the first to be understood and produced by very youngchildren, with the preposition “in” understood before the other two (Bowerman, 1996;Clark, 1973). Recent research with the preferential looking paradigm has demonstratedthat infants as young as 15 months understand these prepositions when shown typicalsituations, as defined by the centrality of the object relative to the reference object. By18 months of age, children have broadened the scope of the meaning of spatial preposi-tions to include less typical locations (Meints, Plunkett, Harris, & Dimmock, 2002). As we have seen, children’s earliest representations of objects, relations, and eventsprovide the universal foundations for linguistic categories. What is the universal con-ceptual knowledge that infants possess about space? In order to map spatial terms to theappropriate referents, infants must attend to one type of relation while ignoring otherpossible relations that may be present in an event. Research over the last several yearshas yielded strong evidence that infants possess sophisticated nonlinguistic knowledgeabout space. As young as 2.5 months, infants understand the basics of the concept ofcontainment, such as the fact that a container must have an opening. By 5 or 6 months,infants also discriminate between the action of placing an object in a tight-fit contain-ment relation and the action of placing an object fitting loosely into a container. They 211. Cognitive Processes in Early Word Learning 203also appreciate that a wide object cannot be placed into a narrower container (Aguiar &Baillargeon, 1998; Spelke & Hespos, 2002). Infants of this age can easily habituate toa container seen from different angles and holding different objects (Casasola, Cohen,& Chiarello, 2003). A few months later, infants as young as 9 months can form a spatialcategory of tight-fit containment events that are different from another category of loose-fit containment events, even if such a distinction is not lexically encoded in their ownlanguage (McDonough, Choi, &[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Mandler, 2003). However, if familiarized with bothtypes of containment, infants can form a more inclusive category of containment (Casa-sola & Cohen, 2002). This precocious understanding of containment explains the resultsof earlier studies that tested the comprehension of spatial prepositions with act- out tasks.These studies demonstrated that children will put objects in a container if there is oneavailable, even if the linguistic instructions require the child to put the object on orunder a container. In addition to the concepts of support and containment, infants also develop knowl-edge about other spatial relations. For example, they possess the categorical representa-tion of “above” versus “below” as early as 3 months of age, though this knowledge doesbecome more robust by 6 months of age (Quinn, Polly, Furer, Dobson, & Narter, 2002).Similarly, infants first demonstrate the ability to discriminate the relation of “between”from a different relation at 6 months of age (Quinn, Adams, Kennedy, Shettler, &Wasnik, 2003). In accord with the proposal that the mapping of spatial terms is not uniquely drivenby cognitive development, a number of recent studies have documented that languagesconflate information about space in different ways into distinct lexical items. For example,English, Spanish, and Dutch use from one to three prepositions to express containment,support, and attachment (Clark, 2004). Studies of the spontaneous speech of childrenhave reported language-related differences as early as 17 to 20 months (see Bowerman& Choi, 2003, for a review). For example, learners of English distinguish betweenactions involving containment (in) and those involving contact/support (on) while learn-ers of Korean ignored this distinction in favor of a categorization of spatial events interms of loose versus tight fitting (kkita vs. nehta). In other words, when the childrentalked about spatial events, they showed sensitivity to language-specific distinctions.This illustrates well the fact that although language acquisition builds on cognitivedevelopment, linguistic representations capture only certain aspects of universal cogni-tive representations as well as emphasizing specific cognitive distinctions not yet repre-sented by the infant.ConclusionsWord learning is a linguistic problem that, unlike many other language acquisition tasks,requires as many tools from the conceptual domain as from the linguistic domain. Aswe have reviewed, infants possess impressive knowledge about object categories andproperties as well as knowledge about relations between entities. They also have devel-oped sophisticated social-cognitive and learning skills that will allow them to establishreference for many types of words. Thus, by the time first words are understood or 212. 204 Diane Poulin-Dubois and Susan A. Grahamproduced, infants have built up strong cognitive functions that they can recruit in theprocess of mapping words to the objects and events. This knowledge base sets the stagefor the prodigious word learning that occurs during the second year of life andbeyond. Throughout the chapter, we have focused on the cognitive underpinnings of wordlearning. At the same time, we must emphasize the critical role that language plays incognition. Over the past decade, numerous[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... studies have demonstrated how linguisticinput shapes cognition. For example, hearing a common name to refer to distinct objectspromotes the formation of object categories in infancy (e.g., Graham et al., 2004;Waxman & Markow, 1995) while hearing distinct names promotes individuation (Xu,2002). Furthermore, the specific language that children are exposed to will also shapethe way they conceptualize the world. For example, researchers have started to examinehow learning language-specific spatial semantic categories influences early spatial seman-tic development (Bowerman & Choi, 2003). In turn, language-specific categories ofspace lead the adult speaker to become increasingly skilled at detecting spatial distinc-tions that are mapped in her mother tongue and less skilled at others (McDonough etal., 2003). When in development this language-specific influence exerts its impact onspatial cognition remains to be determined. The research reviewed in this chapter suggests that infants come well equipped toacquire word meanings, but also that their limited cognitive abilities impact their earlylexical development. That is, recent experimental research on word learning has demon-strated that early in their word learning career, 12- to 13-month-old infants are immatureword learners in many ways. With regard to word reference, infants of that age tend tobe initially biased by superficial cues such as the perceptual salience of objects or actionsinstead of following the perspective of the speaker (Hollich, Hirsh-Pasek, & Golinkoff,2000). They also need more exposure and more cues from the speaker to map a novelword to a referent relative to more experienced word learners (Hirsh-Pasek et al., 2000;Woodward, 2004). Furthermore, at the very early stages of word learning, children havenot yet figured out the ways in which words are different from other symbols (Namy,2001; Woodward & Hoyne, 1999). With regard to word extension, although infantsextend words readily from the beginning, the research that we have reviewed in thischapter indicates that unexperienced word learners are more conservative as well as moreliberal than more experienced word learners in their use of words, regardless of the lin-guistic form class (e.g., Forbes & Farrar, 1993; Meints et al., 1999; Theakston, Lieven,Pine, & Rowland, 2002; Tomasello, 1992). Understanding how these patterns changeover the second year of life and how these changes may be related to cognitive changeswill offer important insights into word learning and contribute to a better understandingof language development in general.NoteThis research was supported by operating grants awarded to both authors by the Natural Sciencesand Engineering Research Council of Canada. Susan Graham also acknowledges support fromthe Canada Research Chairs program. 213. Cognitive Processes in Early Word Learning 205ReferencesAguiar, A., & Baillargeon, R. (1998). Eight-and-a-half-month-old infants’ reasoning about con- tainment events. Child Development, 69, 636–653.Arterberry, M. E., & Bornstein, M. H. (2001). Three-month-old infants’ categorization of animals and vehicles based on static and dynamic attributes. Journal of Experimental Child Psychology, 80, 333–346.Arterberry, M. E., & Bornstein, M. H. (2002).[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Infant perceptual and conceptual categorization: The roles of static and dynamic stimulus attributes. Cognition, 86, 1–24.Au, T. K. F., Dapretto, M., & Song, Y. K. (1994). Input versus constraints: Early word acquisi- tion in Korean and English. Journal of Memory and Language, 33, 567–582.Bahrick, L. E., Gogate, L. J., & Ruiz, I. (2002). Attention and memory for faces and actions in infancy: The salience of actions over faces in dynamic events. Child Development, 73, 1629–1643.Baillargeon, R. (2004). Infants’ reasoning about hidden objects: Evidence for event-general and event-specific expectations. Developmental Science, 7, 391–424.Baker, R. K., Demke, T. L., & Poulin-Dubois, D. (submitted). Infants’ ability to associate motion trajectories with object kinds.Balaban, M. T., & Waxman, S. R. (1997). Do words facilitate object categorization in 9-month- old infants? Journal of Experimental Child Psychology, 64, 3–26.Baldwin, D. A. (1993). Early referential understanding: Infants’ ability to recognize referential acts for what they are. Developmental Psychology, 29, 832–843.Baldwin, D. A., Baird, J. A., Saylor, M. M., & Clark, M. A. (2001). Infants parse dynamic action. Child Development, 72, 708–717.Baldwin, D. A., Markman, E. M., & Melartin, R. L. (1993). Infants’ ability to draw inferences about nonobvious object properties: Evidence from exploratory play. Child Development, 64, 711–728.Barrett, M. D. (1995). Early lexical development. In P. Fletcher & B. MacWhinney (Eds.), Handbook of child language (pp. 362–392). Oxford: Blackwell.Bartsch, K., & Wellman, H. M. (1995). Children talk about the mind. New York: Oxford University Press.Bates, E., Benigni, L., Bretherton, I., Camaioni, L., & Volterra, V. (1979). The emergence of symbols: Cognition and communication in infancy. New York: Academic Press.Bates, E., Marchman, V. A., Thal, D., Fenson, L., Dale, P., & Reznick, S. (1994). Developmental and stylistic variation in the composition of early vocabulary. Journal of Child Language, 21, 85–123.Behne, T., Carpenter, M., Call, J., & Tomasello, M. (2005). Unwilling versus unable: Infants’ understanding of intentional action. Developmental Psychology, 41, 328–337.Bellagamba, F., & Tomasello, M. (1999). Re-enacting intended acts: Comparing 12- and 18- month-olds. Infant Behavior and Development, 22, 277–282.Bloom, P. (1998). Theories of artefact categorization. Cognition, 66, 87–93.Bloom, P. (2000). How children learn the meanings of words. Cambridge, MA: MIT Press.Booth, A. E., & Waxman, S. R. (2002). Object names and object functions serve as cues to categories for infants. Developmental Psychology, 38, 948–957.Bornstein, M. H., Coté, L. R., Maital, S., Painter, K., Park, S. Y., Pascual, L., et al. (2004). Cross-linguistic analysis of vocabulary in young children: Spanish, Dutch, French, Hebrew, Italian, Korean, and American English. Child Development, 75, 1115– 1139. 214. 206 Diane Poulin-Dubois and Susan A. GrahamBowerman, M. (1996). The origins of children’s spatial semantic categories: Cognitive versus linguistic determinants. In J. J. Gumperz & S. C. Levinson (Eds.), Rethinking linguistic relativ- ity (pp. 145–176). New York: Cambridge University Press.Bowerman, M., & Choi, S. (2003). Space under construction:[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Language-specific spatial catego- rization in first language acquisition. In D. Gentner & S. Goldin-Meadow (Eds.), Language in mind: Advances in the study of language and thought (pp. 387–427). Cambridge, MA: MIT Press.Buresh, J. S., Woodward, A., & Brune, C. W. (2006). The roots of verbs in prelinguisitic action knowledge. In K. Hirsh-Pasek & R. M. Golinkoff (Eds.), Action meets word: How children learn verbs (pp. 208–227). New York: Oxford University Press.Carpenter, M., Nagell, K., & Tomasello, M. (1998). Social cognition, joint attention, and com- municative competence from 9 to 15 months of age. Monographs of the Society for Research in Child Development, 63 (4, Serial No. 255).Casasola, M., Bhagwat, J., & Ferguson, K. T. (2006). Precursors to verb learning: Infants’ understanding of motion events. In K. Hirsh-Pasek & R. M. Golinkoff (Eds.), Action meets word: How children learn verbs (pp. 160–190). New York: Oxford University Press.Casasola, M., & Cohen, L. (2002). Infant categorization of containment, support and tight-fit spatial relationships. Developmental Science, 5, 247– 264.Casasola, M., Cohen, L. B., & Chiarello, E. (2003). Six-month-old infants’ categorization of containment spatial relations. Child Development, 74, 679– 693.Caselli, M. C., Bates, E., Casadio, P., Fenson, J., Fenson, L., & Sanderl, L. (1995). A cross-lin- guistic study of early lexical development. Cognitive Development, 10, 159–199.Childers, J. B., & Tomasello, M. (2002). Two- year-olds learn novel nouns, verbs, and conven- tional actions from massed or spaced exposures. Developmental Psychology, 38, 967–978.Choi, S., & Gopnik, A. (1995). Early acquisition of verbs in Korean: A cross-linguistic study. Journal of Child Language, 22, 497–529.Clark, E. V. (1973). What’s in a word? On the child’s acquisition of semantics in his first lan- guage. In T. E. Moore (Ed.), Cognitive development and the acquisition of language. New York: Academic Press.Clark, E. V. (1983). Meanings and concepts. In J. H. Flavell & E. M. Markman (Eds.), Handbook of child development (pp. 787– 840). New York: Wiley.Clark, E. V. (1993). The lexicon in acquisition. Cambridge: Cambridge University Press.Clark, E. V. (2004). How language acquisition builds on cognitive development. Trends in Cogni- tive Sciences, 8, 472–478.Cohen, L. B., & Amsel, G. (1998). Precursors to infants’ perception of the causality of a simple event. Infant Behavior and Development, 21, 713–731.Cohen, L. B., & Oakes, L. M. (1993). How infants perceive a simple causal event. Developmental Psychology, 29, 421– 433.Dromi, E. (1987). Early lexical development. Cambridge: Cambridge University Press.Dromi, E. (1993). The mysteries of early lexical development: Underlying cognitive and linguistic processes in meaning acquisition. In E. Dromi (Ed.), Language and cognition: A developmental perspective (pp. 32–60). Norwood, NJ: Ablex Publishing Co.Fenson, L., Dale, P. S., Reznick, J. S., Bates, E., Thal, D., & Pethick, S. (1994). Variability in early communicative development. Monographs of the Society for Research in Child Development, 59 (Serial No. 242).Forbes, J. N., Ashley, T., & Martin, A. (2003). Interpreting others’ intentions facilitates toddlers’ verb learning. Paper presented at the meeting of the American Psychological Association,[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Atlanta, GA. 215. Cognitive Processes in Early Word Learning 207Forbes, J. N., & Farrar, M. J. (1993). Children’s initial assumptions about the meaning of novel motion verbs: Biased and conservative? Cognitive Development, 8, 273–290.Forbes, J. N., & Farrar, M. J. (1995). Learning to represent word meaning: What initial training events reveal about children’s developing action verb concepts. Cognitive Development, 10, 1–20.Forbes, J. N., & Poulin-Dubois, D. (1997). Representational changes in infants’ interpretation of familiar action word meaning. Journal of Child Language, 24, 389–406.Fox, R., & McDaniel, C. (1982). The perception of biological motion by human infants. Science, 218, 486–487.Gelman, S. A., Croft, W., Fu, P., Clausner, T., & Gottfried, G. (1998). Why is a pomegranate an apple? The role of shape, taxonomic relatedness, and prior lexical knowledge in children’s overextensions of apple and dog. Journal of Child Language, 25, 267–291.Gelman, S. A., & Diesendruck, G. (1999). What’s in a concept? Context, variability, and psy- chological essentialism. In I. E. Sigel (Ed.), Development of mental representation: Theories and applications (pp. 87–111). Mahwah, NJ: Erlbaum.Gentner, D. (1981). Some interesting differences between verbs and nouns. Cognition and Brain Theory, 4, 161–178.Gentner, D. (1982). Why nouns are learned before verbs: Linguistic relativity versus natural partitioning. In S. Kuczaj (Ed.), Language development: Language, thought and culture. Hills- dale, NJ: Erlbaum.Gentner, D., & Boroditsky, L. (2001). Individuation, relativity and early word learning. In M. Bowerman & S. C. Levinson (Eds.), Language acquisition and conceptual development (pp. 215–256). Cambridge: Cambridge University Press.Gershkoff-Stowe, L. (2002). Object naming, vocabulary growth, and the development of word retrieval abilities. Journal of Memory and Language, 46, 665–687.Gershkoff-Stowe, L., & Smith, L. B. (2004). Shape and the first hundred nouns. Child Develop- ment, 75, 1098– 1114.Gershkoff-Stowe, L., Thal, D. J., Smith, L. B., & Namy, L. L. (1997). Categorization and its developmental relation to early language. Child Development, 69, 843–859.Golinkoff, R. M., Chung, H. L., Hirsh-Pasek, K., Liu, J., Bertenthal, B. I., Brand, R., et al. (2002). Young children can extend motion verbs to point-light displays. Developmental Psychol- ogy, 38, 604– 614.Gopnik, A. (1984). The acquisition of gone and the development of the object concept. Journal of Child Language, 11, 273–292.Gopnik, A., & Meltzoff, A. N. (1986). Relations between semantic and cognitive develop- ment in the one-word stage: The specificity hypothesis. Child Development, 57, 1040– 1053.Gopnik, A., & Meltzoff, A. (1987). The development of categorization in the second year and its relation to other cognitive and linguistic developments. Child Development, 58, 1523–1531.Graham, S. A., Kilbreath, C. S., & Welder, A. N. (2004). 13-month-olds rely on shared labels and shape similarity for inductive inferences. Child Development, 75, 409– 427.Graham, S. A., & Poulin-Dubois, D. (1999). Infants’ use of shape to extend novel labels to animate and inanimate objects. Journal of Child Language, 26, 295–320.Hall, D. G., & Waxman, S. R. (Eds.). (2004).[9/19/2012 2:58:17 PM]
    • Blackwell Handbooks of Developmenal Psychology Erika Hoff Marilyn S... Weaving a lexicon. Cambridge, MA: MIT Press.Hirsh-Pasek, K., Golinkoff, R. M., & Hollich, G. (2000). An emergentist coalition model for word learning: Mapping words to objects is a product of the interaction of multiple cues. In 216. 208 Diane Poulin-Dubois and Susan A. Graham R. M. Golinkoff, K. Hirsh-Pasek, L. Bloom, L. B. Smith, A. L. Woodward, N. Akhtar, et al. (Eds.), Becoming a word learner: A debate on lexical acquisition (pp. 136–164). Oxford: Oxford University Press.Hollich, G. J., Hirsh-Pasek, K., & Golinkoff, R. M. (2000). Breaking the language barrier: An emergent coalition model for the origins of word learning. Monographs of the Society for Research in Child Development, 65 (3, Serial No. 262).Huttenlocher, J. (1974). The origin of language comprehension. In R. L. Solso (Ed.), Theories of cognitive psychology: The Loyola Symposium (pp. 331–368). New York: Halsted Press, Winston-Wiley.Imai, M., Haryu, E., & Okada, H. (2005). Mapping novel nouns and verbs onto dynamic action events: Are verb meanings easier to learn than noun meanings for Japanese children? Child Development, 76, 340– 355.Jackson-Maldonado, D., Thal, D., Marchman, V., Bates, E., & Gutierrez- Clellen, V. (1993). Early lexical development in Spanish-speaking infants and toddlers. Journal of Child Language, 20, 523–549.Johnson, S. P. (2004). Development of perceptual completion in infancy. Psychological Science, 15, 769–775.Johnson, S. P., & Aslin, R. N. (1995). Perception of object unity in 2-month-old i